OPTION VALUES: Model directory: ./IOE2+0.07/ Training data file: train.tagged Testing data file: test.tagged Unlabeled data file: data.untagged Model file: model.txt Training log file (this one): trainlog.txt Label representation: IOE2 Second-order Markov CRFs Number of labels: 24 Number of training sequences: 443 (one data partition) Number of testing sequences: 22 (one data partition) Number of unlabeled sequences: 0 Number of context predicates: 687508 Number of features: 1466312 Feature rare threshold: 1 Context predicate rare threshold: 1 Using multiple rare thresholds for features: 0 Highlight feature: 0 Number of training iterations: 200 Initial lambda value: 0.0700 Sigma square (for smoothing): 100.0000 Epsilon for L-BFGS convergence: 0.000100 Number of approximated hessian matrixes: 7 Start to train ... Iteration: 1 Log-likelihood = -4316281.806510 Norm (log-likelihood gradient vector) = 511317.208310 Norm (lambda vector) = 84.763959 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 320 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4802 4509 93.90 97.32 95.58 i-np 13660 15067 13293 88.23 97.31 92.55 e-np 12220 12321 11630 94.39 95.17 94.78 o 6349 6001 5791 96.50 91.21 93.78 e-vp 4768 4575 4441 97.07 93.14 95.07 i-vp 2602 2518 2405 95.51 92.43 93.95 e-adjp 384 159 151 94.97 39.32 55.62 i-pp 52 26 25 96.15 48.08 64.10 e-advp 822 534 484 90.64 58.88 71.39 i-advp 100 22 21 95.45 21.00 34.43 e-sbar 503 274 268 97.81 53.28 68.98 i-adjp 152 28 25 89.29 16.45 27.78 e-prt 126 114 109 95.61 86.51 90.83 i-sbar 12 1 1 100.00 8.33 15.38 i-conjp 24 1 1 100.00 4.17 8.00 e-conjp 16 1 1 100.00 6.25 11.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 7 7 100.00 70.00 82.35 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.28 48.94 61.10 Avg2. 46451 46451 43162 92.92 92.92 92.92 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12333 10495 85.10 85.88 85.49 pp 4633 4802 4495 93.61 97.02 95.28 vp 4768 4575 4294 93.86 90.06 91.92 sbar 503 274 263 95.99 52.29 67.70 adjp 384 160 147 91.88 38.28 54.04 advp 822 534 483 90.45 58.76 71.24 prt 126 114 109 95.61 86.51 90.83 lst 10 7 7 100.00 70.00 82.35 intj 4 0 0 0.00 0.00 0.00 conjp 16 1 1 100.00 6.25 11.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 84.65 58.50 69.19 Avg2. 23486 22800 20294 89.01 86.41 87.69 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 354 seconds Iteration: 2 Log-likelihood = -3817090.852107 Norm (log-likelihood gradient vector) = 485308.116113 Norm (lambda vector) = 84.806104 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4829 4027 83.39 86.92 85.12 i-np 13660 20246 13268 65.53 97.13 78.26 e-np 12220 12182 10122 83.09 82.83 82.96 o 6349 5318 4896 92.06 77.11 83.93 e-vp 4768 2559 2221 86.79 46.58 60.63 i-vp 2602 1298 1276 98.31 49.04 65.44 e-adjp 384 7 7 100.00 1.82 3.58 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 11 11 100.00 1.34 2.64 i-advp 100 0 0 0.00 0.00 0.00 e-sbar 503 0 0 0.00 0.00 0.00 i-adjp 152 0 0 0.00 0.00 0.00 e-prt 126 1 1 100.00 0.79 1.57 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 40.46 22.18 28.65 Avg2. 46451 46451 35829 77.13 77.13 77.13 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12189 6981 57.27 57.13 57.20 pp 4633 4829 3998 82.79 86.29 84.51 vp 4768 2559 2135 83.43 44.78 58.28 sbar 503 0 0 0.00 0.00 0.00 adjp 384 7 7 100.00 1.82 3.58 advp 822 11 11 100.00 1.34 2.64 prt 126 1 1 100.00 0.79 1.57 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 52.35 19.22 28.11 Avg2. 23486 19596 13133 67.02 55.92 60.97 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 3 Log-likelihood = -2282963.871987 Norm (log-likelihood gradient vector) = 280793.670898 Norm (lambda vector) = 85.092075 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4767 4022 84.37 86.81 85.57 i-np 13660 22023 13325 60.50 97.55 74.69 e-np 12220 11250 8864 78.79 72.54 75.53 o 6349 5174 4819 93.14 75.90 83.64 e-vp 4768 2142 1813 84.64 38.02 52.47 i-vp 2602 1095 1078 98.45 41.43 58.32 e-adjp 384 0 0 0.00 0.00 0.00 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 0 0 0.00 0.00 0.00 i-advp 100 0 0 0.00 0.00 0.00 e-sbar 503 0 0 0.00 0.00 0.00 i-adjp 152 0 0 0.00 0.00 0.00 e-prt 126 0 0 0.00 0.00 0.00 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 24.99 20.61 22.59 Avg2. 46451 46451 33921 73.03 73.03 73.03 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 11255 5748 51.07 47.04 48.97 pp 4633 4767 3991 83.72 86.14 84.91 vp 4768 2142 1678 78.34 35.19 48.57 sbar 503 0 0 0.00 0.00 0.00 adjp 384 0 0 0.00 0.00 0.00 advp 822 0 0 0.00 0.00 0.00 prt 126 0 0 0.00 0.00 0.00 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 21.31 16.84 18.81 Avg2. 23486 18164 11417 62.86 48.61 54.82 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 4 Log-likelihood = -1550905.513473 Norm (log-likelihood gradient vector) = 257367.056752 Norm (lambda vector) = 86.148535 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4967 4327 87.11 93.40 90.15 i-np 13660 15930 12932 81.18 94.67 87.41 e-np 12220 13138 11104 84.52 90.87 87.58 o 6349 6070 5630 92.75 88.68 90.67 e-vp 4768 4173 3704 88.76 77.68 82.85 i-vp 2602 2173 2003 92.18 76.98 83.90 e-adjp 384 0 0 0.00 0.00 0.00 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 0 0 0.00 0.00 0.00 i-advp 100 0 0 0.00 0.00 0.00 e-sbar 503 0 0 0.00 0.00 0.00 i-adjp 152 0 0 0.00 0.00 0.00 e-prt 126 0 0 0.00 0.00 0.00 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 26.33 26.11 26.22 Avg2. 46451 46451 39700 85.47 85.47 85.47 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 13145 9340 71.05 76.43 73.64 pp 4633 4967 4291 86.39 92.62 89.40 vp 4768 4173 3480 83.39 72.99 77.84 sbar 503 0 0 0.00 0.00 0.00 adjp 384 0 0 0.00 0.00 0.00 advp 822 0 0 0.00 0.00 0.00 prt 126 0 0 0.00 0.00 0.00 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 24.08 24.20 24.14 Avg2. 23486 22285 17111 76.78 72.86 74.77 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 5 Log-likelihood = -1200172.472800 Norm (log-likelihood gradient vector) = 135185.722760 Norm (lambda vector) = 86.427988 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 5104 4441 87.01 95.86 91.22 i-np 13660 14719 12940 87.91 94.73 91.19 e-np 12220 13061 11523 88.22 94.30 91.16 o 6349 6028 5702 94.59 89.81 92.14 e-vp 4768 4742 4182 88.19 87.71 87.95 i-vp 2602 2797 2345 83.84 90.12 86.87 e-adjp 384 0 0 0.00 0.00 0.00 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 0 0 0.00 0.00 0.00 i-advp 100 0 0 0.00 0.00 0.00 e-sbar 503 0 0 0.00 0.00 0.00 i-adjp 152 0 0 0.00 0.00 0.00 e-prt 126 0 0 0.00 0.00 0.00 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 26.49 27.63 27.05 Avg2. 46451 46451 41133 88.55 88.55 88.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 13063 10263 78.57 83.99 81.18 pp 4633 5104 4401 86.23 94.99 90.40 vp 4768 4742 3871 81.63 81.19 81.41 sbar 503 0 0 0.00 0.00 0.00 adjp 384 0 0 0.00 0.00 0.00 advp 822 0 0 0.00 0.00 0.00 prt 126 0 0 0.00 0.00 0.00 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 24.64 26.02 25.31 Avg2. 23486 22909 18535 80.91 78.92 79.90 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 6 Log-likelihood = -998345.393933 Norm (log-likelihood gradient vector) = 81154.948757 Norm (lambda vector) = 86.752458 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 5128 4474 87.25 96.57 91.67 i-np 13660 14040 12954 92.26 94.83 93.53 e-np 12220 12811 11784 91.98 96.43 94.16 o 6349 6036 5740 95.10 90.41 92.69 e-vp 4768 4957 4261 85.96 89.37 87.63 i-vp 2602 3313 2483 74.95 95.43 83.96 e-adjp 384 0 0 0.00 0.00 0.00 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 148 135 91.22 16.42 27.84 i-advp 100 0 0 0.00 0.00 0.00 e-sbar 503 18 18 100.00 3.58 6.91 i-adjp 152 0 0 0.00 0.00 0.00 e-prt 126 0 0 0.00 0.00 0.00 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 35.94 29.15 32.19 Avg2. 46451 46451 41849 90.09 90.09 90.09 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12812 10826 84.50 88.59 86.50 pp 4633 5128 4434 86.47 95.70 90.85 vp 4768 4957 3860 77.87 80.96 79.38 sbar 503 18 18 100.00 3.58 6.91 adjp 384 0 0 0.00 0.00 0.00 advp 822 148 133 89.86 16.18 27.42 prt 126 0 0 0.00 0.00 0.00 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 43.87 28.50 34.55 Avg2. 23486 23063 19271 83.56 82.05 82.80 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 7 Log-likelihood = -841455.061326 Norm (log-likelihood gradient vector) = 66564.557333 Norm (lambda vector) = 87.305538 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4805 4482 93.28 96.74 94.98 i-np 13660 13882 13083 94.24 95.78 95.00 e-np 12220 12497 11856 94.87 97.02 95.93 o 6349 6128 5868 95.76 92.42 94.06 e-vp 4768 5018 4474 89.16 93.83 91.44 i-vp 2602 3177 2520 79.32 96.85 87.21 e-adjp 384 154 148 96.10 38.54 55.02 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 483 395 81.78 48.05 60.54 i-advp 100 0 0 0.00 0.00 0.00 e-sbar 503 297 283 95.29 56.26 70.75 i-adjp 152 8 8 100.00 5.26 10.00 e-prt 126 2 2 100.00 1.59 3.12 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 50.99 36.12 42.28 Avg2. 46451 46451 43119 92.83 92.83 92.83 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12500 11123 88.98 91.02 89.99 pp 4633 4805 4445 92.51 95.94 94.19 vp 4768 5019 4128 82.25 86.58 84.36 sbar 503 297 277 93.27 55.07 69.25 adjp 384 154 124 80.52 32.29 46.10 advp 822 483 354 73.29 43.07 54.25 prt 126 2 2 100.00 1.59 3.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 61.08 40.56 48.75 Avg2. 23486 23260 20453 87.93 87.09 87.51 Current max chunk-based F1: 87.69 (iteration 1) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 8 Log-likelihood = -675379.114644 Norm (log-likelihood gradient vector) = 70842.545105 Norm (lambda vector) = 88.758543 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4817 4501 93.44 97.15 95.26 i-np 13660 13414 12909 96.24 94.50 95.36 e-np 12220 12528 11942 95.32 97.73 96.51 o 6349 6460 6096 94.37 96.02 95.18 e-vp 4768 4892 4541 92.83 95.24 94.02 i-vp 2602 2898 2502 86.34 96.16 90.98 e-adjp 384 306 252 82.35 65.62 73.04 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 730 562 76.99 68.37 72.42 i-advp 100 1 1 100.00 1.00 1.98 e-sbar 503 339 315 92.92 62.62 74.82 i-adjp 152 36 29 80.56 19.08 30.85 e-prt 126 30 30 100.00 23.81 38.46 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 54.57 40.86 46.73 Avg2. 46451 46451 43680 94.03 94.03 94.03 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12537 11261 89.82 92.15 90.97 pp 4633 4817 4462 92.63 96.31 94.43 vp 4768 4893 4307 88.02 90.33 89.16 sbar 503 339 310 91.45 61.63 73.63 adjp 384 306 204 66.67 53.12 59.13 advp 822 730 515 70.55 62.65 66.37 prt 126 30 30 100.00 23.81 38.46 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 59.91 48.00 53.30 Avg2. 23486 23652 21089 89.16 89.79 89.48 Current max chunk-based F1: 89.48 (iteration 8) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 9 Log-likelihood = -563456.319949 Norm (log-likelihood gradient vector) = 44754.658783 Norm (lambda vector) = 90.387569 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4849 4531 93.44 97.80 95.57 i-np 13660 13709 13111 95.64 95.98 95.81 e-np 12220 12277 11879 96.76 97.21 96.98 o 6349 6340 6073 95.79 95.65 95.72 e-vp 4768 4852 4569 94.17 95.83 94.99 i-vp 2602 2803 2498 89.12 96.00 92.43 e-adjp 384 355 281 79.15 73.18 76.05 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 803 605 75.34 73.60 74.46 i-advp 100 5 4 80.00 4.00 7.62 e-sbar 503 338 313 92.60 62.23 74.44 i-adjp 152 57 41 71.93 26.97 39.23 e-prt 126 63 62 98.41 49.21 65.61 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 53.12 43.38 47.76 Avg2. 46451 46451 43967 94.65 94.65 94.65 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12288 11233 91.41 91.92 91.67 pp 4633 4849 4492 92.64 96.96 94.75 vp 4768 4852 4364 89.94 91.53 90.73 sbar 503 338 308 91.12 61.23 73.25 adjp 384 355 228 64.23 59.38 61.71 advp 822 803 557 69.36 67.76 68.55 prt 126 63 62 98.41 49.21 65.61 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 59.71 51.80 55.47 Avg2. 23486 23548 21244 90.22 90.45 90.33 Current max chunk-based F1: 90.33 (iteration 9) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 10 Log-likelihood = -498082.539553 Norm (log-likelihood gradient vector) = 26529.961713 Norm (lambda vector) = 91.829463 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4805 4545 94.59 98.10 96.31 i-np 13660 13615 13125 96.40 96.08 96.24 e-np 12220 12268 11907 97.06 97.44 97.25 o 6349 6388 6113 95.70 96.28 95.99 e-vp 4768 4852 4608 94.97 96.64 95.80 i-vp 2602 2710 2481 91.55 95.35 93.41 e-adjp 384 401 304 75.81 79.17 77.45 i-pp 52 0 0 0.00 0.00 0.00 e-advp 822 819 632 77.17 76.89 77.03 i-advp 100 17 15 88.24 15.00 25.64 e-sbar 503 391 358 91.56 71.17 80.09 i-adjp 152 93 64 68.82 42.11 52.24 e-prt 126 92 90 97.83 71.43 82.57 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 53.48 46.78 49.91 Avg2. 46451 46451 44242 95.24 95.24 95.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12277 11331 92.29 92.73 92.51 pp 4633 4805 4506 93.78 97.26 95.49 vp 4768 4852 4420 91.10 92.70 91.89 sbar 503 391 352 90.03 69.98 78.75 adjp 384 401 256 63.84 66.67 65.22 advp 822 819 595 72.65 72.38 72.52 prt 126 92 90 97.83 71.43 82.57 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 60.15 56.31 58.17 Avg2. 23486 23637 21550 91.17 91.76 91.46 Current max chunk-based F1: 91.46 (iteration 10) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 11 Log-likelihood = -446609.835349 Norm (log-likelihood gradient vector) = 22181.786495 Norm (lambda vector) = 94.391101 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4823 4551 94.36 98.23 96.26 i-np 13660 14536 13381 92.05 97.96 94.91 e-np 12220 11776 11565 98.21 94.64 96.39 o 6349 5959 5808 97.47 91.48 94.38 e-vp 4768 4817 4617 95.85 96.83 96.34 i-vp 2602 2669 2477 92.81 95.20 93.99 e-adjp 384 398 305 76.63 79.43 78.01 i-pp 52 10 10 100.00 19.23 32.26 e-advp 822 808 647 80.07 78.71 79.39 i-advp 100 57 39 68.42 39.00 49.68 e-sbar 503 378 342 90.48 67.99 77.64 i-adjp 152 109 76 69.72 50.00 58.24 e-prt 126 111 108 97.30 85.71 91.14 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 57.67 49.72 53.40 Avg2. 46451 46451 43926 94.56 94.56 94.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 11792 10874 92.22 88.99 90.57 pp 4633 4823 4522 93.76 97.60 95.64 vp 4768 4817 4443 92.24 93.18 92.71 sbar 503 378 336 88.89 66.80 76.28 adjp 384 398 266 66.83 69.27 68.03 advp 822 808 627 77.60 76.28 76.93 prt 126 111 108 97.30 85.71 91.14 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 60.88 57.78 59.29 Avg2. 23486 23127 21176 91.56 90.16 90.86 Current max chunk-based F1: 91.46 (iteration 10) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 12 Log-likelihood = -439461.694185 Norm (log-likelihood gradient vector) = 61526.457274 Norm (lambda vector) = 98.880588 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4823 4555 94.44 98.32 96.34 i-np 13660 13685 13197 96.43 96.61 96.52 e-np 12220 12208 11895 97.44 97.34 97.39 o 6349 6358 6110 96.10 96.24 96.17 e-vp 4768 4823 4627 95.94 97.04 96.49 i-vp 2602 2661 2479 93.16 95.27 94.20 e-adjp 384 399 309 77.44 80.47 78.93 i-pp 52 19 19 100.00 36.54 53.52 e-advp 822 802 647 80.67 78.71 79.68 i-advp 100 61 41 67.21 41.00 50.93 e-sbar 503 377 344 91.25 68.39 78.18 i-adjp 152 119 82 68.91 53.95 60.52 e-prt 126 116 112 96.55 88.89 92.56 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 57.78 51.44 54.42 Avg2. 46451 46451 44417 95.62 95.62 95.62 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12220 11380 93.13 93.13 93.13 pp 4633 4823 4535 94.03 97.88 95.92 vp 4768 4823 4460 92.47 93.54 93.00 sbar 503 377 337 89.39 67.00 76.59 adjp 384 399 272 68.17 70.83 69.48 advp 822 802 629 78.43 76.52 77.46 prt 126 116 112 96.55 88.89 92.56 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 61.22 58.78 59.97 Avg2. 23486 23560 21725 92.21 92.50 92.36 Current max chunk-based F1: 92.36 (iteration 12) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 13 Log-likelihood = -396824.644723 Norm (log-likelihood gradient vector) = 21379.458325 Norm (lambda vector) = 100.257763 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4817 4557 94.60 98.36 96.44 i-np 13660 13574 13163 96.97 96.36 96.67 e-np 12220 12285 11938 97.18 97.69 97.43 o 6349 6419 6148 95.78 96.83 96.30 e-vp 4768 4819 4627 96.02 97.04 96.53 i-vp 2602 2670 2485 93.07 95.50 94.27 e-adjp 384 383 304 79.37 79.17 79.27 i-pp 52 26 26 100.00 50.00 66.67 e-advp 822 786 639 81.30 77.74 79.48 i-advp 100 64 43 67.19 43.00 52.44 e-sbar 503 373 344 92.23 68.39 78.54 i-adjp 152 117 82 70.09 53.95 60.97 e-prt 126 118 112 94.92 88.89 91.80 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 57.93 52.15 54.89 Avg2. 46451 46451 44468 95.73 95.73 95.73 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12291 11444 93.11 93.65 93.38 pp 4633 4817 4544 94.33 98.08 96.17 vp 4768 4819 4459 92.53 93.52 93.02 sbar 503 373 337 90.35 67.00 76.94 adjp 384 383 268 69.97 69.79 69.88 advp 822 786 622 79.13 75.67 77.36 prt 126 118 112 94.92 88.89 91.80 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 61.43 58.66 60.01 Avg2. 23486 23587 21786 92.36 92.76 92.56 Current max chunk-based F1: 92.56 (iteration 13) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 14 Log-likelihood = -383734.505439 Norm (log-likelihood gradient vector) = 17720.625614 Norm (lambda vector) = 100.628294 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4768 4557 95.57 98.36 96.95 i-np 13660 13441 13095 97.43 95.86 96.64 e-np 12220 12381 11982 96.78 98.05 97.41 o 6349 6489 6175 95.16 97.26 96.20 e-vp 4768 4809 4631 96.30 97.13 96.71 i-vp 2602 2669 2491 93.33 95.73 94.52 e-adjp 384 365 296 81.10 77.08 79.04 i-pp 52 32 32 100.00 61.54 76.19 e-advp 822 779 645 82.80 78.47 80.57 i-advp 100 74 48 64.86 48.00 55.17 e-sbar 503 403 375 93.05 74.55 82.78 i-adjp 152 117 82 70.09 53.95 60.97 e-prt 126 124 117 94.35 92.86 93.60 i-sbar 12 0 0 0.00 0.00 0.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 58.04 53.44 55.65 Avg2. 46451 46451 44526 95.86 95.86 95.86 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12383 11508 92.93 94.17 93.55 pp 4633 4768 4549 95.41 98.19 96.78 vp 4768 4809 4476 93.08 93.88 93.47 sbar 503 403 365 90.57 72.56 80.57 adjp 384 365 265 72.60 69.01 70.76 advp 822 779 629 80.74 76.52 78.58 prt 126 124 117 94.35 92.86 93.60 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 61.97 59.72 60.82 Avg2. 23486 23631 21909 92.71 93.29 93.00 Current max chunk-based F1: 93.00 (iteration 14) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 15 Log-likelihood = -360413.567329 Norm (log-likelihood gradient vector) = 21642.779964 Norm (lambda vector) = 102.334087 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4724 4546 96.23 98.12 97.17 i-np 13660 13564 13165 97.06 96.38 96.72 e-np 12220 12327 11963 97.05 97.90 97.47 o 6349 6417 6145 95.76 96.79 96.27 e-vp 4768 4793 4632 96.64 97.15 96.89 i-vp 2602 2667 2492 93.44 95.77 94.59 e-adjp 384 349 293 83.95 76.30 79.95 i-pp 52 38 33 86.84 63.46 73.33 e-advp 822 795 661 83.14 80.41 81.76 i-advp 100 82 52 63.41 52.00 57.14 e-sbar 503 444 403 90.77 80.12 85.11 i-adjp 152 122 95 77.87 62.50 69.34 e-prt 126 127 120 94.49 95.24 94.86 i-sbar 12 2 1 50.00 8.33 14.29 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 60.33 55.02 57.56 Avg2. 46451 46451 44601 96.02 96.02 96.02 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12327 11504 93.32 94.14 93.73 pp 4633 4724 4534 95.98 97.86 96.91 vp 4768 4793 4481 93.49 93.98 93.73 sbar 503 444 391 88.06 77.73 82.58 adjp 384 349 271 77.65 70.57 73.94 advp 822 795 645 81.13 78.47 79.78 prt 126 127 120 94.49 95.24 94.86 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 62.41 60.80 61.60 Avg2. 23486 23559 21946 93.15 93.44 93.30 Current max chunk-based F1: 93.30 (iteration 15) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 16 Log-likelihood = -340065.209039 Norm (log-likelihood gradient vector) = 19434.398958 Norm (lambda vector) = 104.612059 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4715 4554 96.59 98.29 97.43 i-np 13660 13452 13135 97.64 96.16 96.89 e-np 12220 12366 11994 96.99 98.15 97.57 o 6349 6444 6171 95.76 97.20 96.47 e-vp 4768 4767 4632 97.17 97.15 97.16 i-vp 2602 2689 2503 93.08 96.20 94.61 e-adjp 384 343 291 84.84 75.78 80.06 i-pp 52 46 34 73.91 65.38 69.39 e-advp 822 821 669 81.49 81.39 81.44 i-advp 100 93 60 64.52 60.00 62.18 e-sbar 503 444 407 91.67 80.91 85.96 i-adjp 152 130 108 83.08 71.05 76.60 e-prt 126 128 120 93.75 95.24 94.49 i-sbar 12 4 2 50.00 16.67 25.00 i-conjp 24 6 6 100.00 25.00 40.00 e-conjp 16 3 3 100.00 18.75 31.58 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.02 58.67 63.84 Avg2. 46451 46451 44689 96.21 96.21 96.21 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12367 11563 93.50 94.62 94.06 pp 4633 4715 4536 96.20 97.91 97.05 vp 4768 4767 4481 94.00 93.98 93.99 sbar 503 444 395 88.96 78.53 83.42 adjp 384 343 276 80.47 71.88 75.93 advp 822 821 654 79.66 79.56 79.61 prt 126 128 120 93.75 95.24 94.49 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 3 3 100.00 18.75 31.58 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.65 63.05 67.51 Avg2. 23486 23588 22028 93.39 93.79 93.59 Current max chunk-based F1: 93.59 (iteration 16) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 17 Log-likelihood = -325220.463919 Norm (log-likelihood gradient vector) = 19414.760648 Norm (lambda vector) = 108.126316 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4714 4557 96.67 98.36 97.51 i-np 13660 13727 13306 96.93 97.41 97.17 e-np 12220 12203 11930 97.76 97.63 97.69 o 6349 6291 6113 97.17 96.28 96.72 e-vp 4768 4762 4636 97.35 97.23 97.29 i-vp 2602 2700 2516 93.19 96.69 94.91 e-adjp 384 348 296 85.06 77.08 80.87 i-pp 52 42 33 78.57 63.46 70.21 e-advp 822 853 690 80.89 83.94 82.39 i-advp 100 90 60 66.67 60.00 63.16 e-sbar 503 448 411 91.74 81.71 86.44 i-adjp 152 126 107 84.92 70.39 76.98 e-prt 126 129 120 93.02 95.24 94.12 i-sbar 12 4 2 50.00 16.67 25.00 i-conjp 24 10 10 100.00 41.67 58.82 e-conjp 16 4 4 100.00 25.00 40.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.50 59.94 64.79 Avg2. 46451 46451 44791 96.43 96.43 96.43 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12204 11519 94.39 94.26 94.33 pp 4633 4714 4541 96.33 98.01 97.16 vp 4768 4762 4488 94.25 94.13 94.19 sbar 503 448 399 89.06 79.32 83.91 adjp 384 348 282 81.03 73.44 77.05 advp 822 853 675 79.13 82.12 80.60 prt 126 129 120 93.02 95.24 94.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 5 4 80.00 25.00 38.10 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.72 64.15 67.28 Avg2. 23486 23463 22028 93.88 93.79 93.84 Current max chunk-based F1: 93.84 (iteration 17) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 18 Log-likelihood = -307760.651922 Norm (log-likelihood gradient vector) = 13642.792979 Norm (lambda vector) = 109.967807 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4720 4563 96.67 98.49 97.57 i-np 13660 13850 13351 96.40 97.74 97.06 e-np 12220 12119 11896 98.16 97.35 97.75 o 6349 6230 6080 97.59 95.76 96.67 e-vp 4768 4760 4629 97.25 97.08 97.17 i-vp 2602 2707 2517 92.98 96.73 94.82 e-adjp 384 360 303 84.17 78.91 81.45 i-pp 52 38 33 86.84 63.46 73.33 e-advp 822 861 698 81.07 84.91 82.95 i-advp 100 88 59 67.05 59.00 62.77 e-sbar 503 441 412 93.42 81.91 87.29 i-adjp 152 127 107 84.25 70.39 76.70 e-prt 126 129 120 93.02 95.24 94.12 i-sbar 12 4 2 50.00 16.67 25.00 i-conjp 24 12 12 100.00 50.00 66.67 e-conjp 16 5 5 100.00 31.25 47.62 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.94 60.74 65.45 Avg2. 46451 46451 44787 96.42 96.42 96.42 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12120 11474 94.67 93.90 94.28 pp 4633 4720 4553 96.46 98.27 97.36 vp 4768 4760 4477 94.05 93.90 93.98 sbar 503 441 400 90.70 79.52 84.75 adjp 384 360 289 80.28 75.26 77.69 advp 822 861 682 79.21 82.97 81.05 prt 126 129 120 93.02 95.24 94.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 6 5 83.33 31.25 45.45 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.17 65.03 67.96 Avg2. 23486 23397 22000 94.03 93.67 93.85 Current max chunk-based F1: 93.85 (iteration 18) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 19 Log-likelihood = -294928.586863 Norm (log-likelihood gradient vector) = 14603.992122 Norm (lambda vector) = 111.427194 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4696 4560 97.10 98.42 97.76 i-np 13660 13799 13341 96.68 97.66 97.17 e-np 12220 12129 11916 98.24 97.51 97.88 o 6349 6253 6101 97.57 96.09 96.83 e-vp 4768 4771 4642 97.30 97.36 97.33 i-vp 2602 2709 2528 93.32 97.16 95.20 e-adjp 384 378 314 83.07 81.77 82.41 i-pp 52 35 33 94.29 63.46 75.86 e-advp 822 840 695 82.74 84.55 83.63 i-advp 100 82 60 73.17 60.00 65.93 e-sbar 503 465 435 93.55 86.48 89.88 i-adjp 152 134 113 84.33 74.34 79.02 e-prt 126 129 120 93.02 95.24 94.12 i-sbar 12 5 3 60.00 25.00 35.29 i-conjp 24 18 16 88.89 66.67 76.19 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.66 63.59 67.38 Avg2. 46451 46451 44885 96.63 96.63 96.63 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12129 11516 94.95 94.24 94.59 pp 4633 4696 4552 96.93 98.25 97.59 vp 4768 4771 4499 94.30 94.36 94.33 sbar 503 465 424 91.18 84.29 87.60 adjp 384 378 300 79.37 78.12 78.74 advp 822 840 682 81.19 82.97 82.07 prt 126 129 120 93.02 95.24 94.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.98 67.75 69.80 Avg2. 23486 23417 22101 94.38 94.10 94.24 Current max chunk-based F1: 94.24 (iteration 19) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 20 Log-likelihood = -277176.095397 Norm (log-likelihood gradient vector) = 12862.273126 Norm (lambda vector) = 114.941803 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4725 4571 96.74 98.66 97.69 i-np 13660 13702 13313 97.16 97.46 97.31 e-np 12220 12216 11974 98.02 97.99 98.00 o 6349 6317 6144 97.26 96.77 97.02 e-vp 4768 4766 4642 97.40 97.36 97.38 i-vp 2602 2712 2534 93.44 97.39 95.37 e-adjp 384 373 314 84.18 81.77 82.96 i-pp 52 34 33 97.06 63.46 76.74 e-advp 822 785 677 86.24 82.36 84.26 i-advp 100 73 58 79.45 58.00 67.05 e-sbar 503 450 428 95.11 85.09 89.82 i-adjp 152 133 112 84.21 73.68 78.60 e-prt 126 129 121 93.80 96.03 94.90 i-sbar 12 7 5 71.43 41.67 52.63 i-conjp 24 20 18 90.00 75.00 81.82 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 73.08 64.95 68.77 Avg2. 46451 46451 44953 96.78 96.78 96.78 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12216 11592 94.89 94.86 94.88 pp 4633 4725 4564 96.59 98.51 97.54 vp 4768 4766 4502 94.46 94.42 94.44 sbar 503 450 419 93.11 83.30 87.93 adjp 384 373 301 80.70 78.39 79.52 advp 822 785 664 84.59 80.78 82.64 prt 126 129 121 93.80 96.03 94.90 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.81 68.25 70.46 Avg2. 23486 23454 22172 94.53 94.41 94.47 Current max chunk-based F1: 94.47 (iteration 20) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 21 Log-likelihood = -256406.553028 Norm (log-likelihood gradient vector) = 12545.309445 Norm (lambda vector) = 120.673937 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4697 4563 97.15 98.49 97.81 i-np 13660 13702 13305 97.10 97.40 97.25 e-np 12220 12211 11962 97.96 97.89 97.92 o 6349 6332 6149 97.11 96.85 96.98 e-vp 4768 4780 4648 97.24 97.48 97.36 i-vp 2602 2708 2529 93.39 97.19 95.25 e-adjp 384 373 316 84.72 82.29 83.49 i-pp 52 34 33 97.06 63.46 76.74 e-advp 822 769 664 86.35 80.78 83.47 i-advp 100 71 56 78.87 56.00 65.50 e-sbar 503 469 442 94.24 87.87 90.95 i-adjp 152 136 114 83.82 75.00 79.17 e-prt 126 129 121 93.80 96.03 94.90 i-sbar 12 10 6 60.00 50.00 54.55 i-conjp 24 21 18 85.71 75.00 80.00 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.23 65.40 68.64 Avg2. 46451 46451 44935 96.74 96.74 96.74 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12211 11581 94.84 94.77 94.81 pp 4633 4697 4556 97.00 98.34 97.66 vp 4768 4780 4505 94.25 94.48 94.37 sbar 503 469 432 92.11 85.88 88.89 adjp 384 373 305 81.77 79.43 80.58 advp 822 769 651 84.66 79.20 81.84 prt 126 129 121 93.80 96.03 94.90 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 11 9 81.82 56.25 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.02 68.44 70.19 Avg2. 23486 23439 22160 94.54 94.35 94.45 Current max chunk-based F1: 94.47 (iteration 20) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 22 Log-likelihood = -247956.345617 Norm (log-likelihood gradient vector) = 11307.484014 Norm (lambda vector) = 123.136194 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4675 4553 97.39 98.27 97.83 i-np 13660 13570 13253 97.66 97.02 97.34 e-np 12220 12281 12001 97.72 98.21 97.96 o 6349 6384 6168 96.62 97.15 96.88 e-vp 4768 4776 4650 97.36 97.53 97.44 i-vp 2602 2703 2526 93.45 97.08 95.23 e-adjp 384 375 319 85.07 83.07 84.06 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 779 672 86.26 81.75 83.95 i-advp 100 73 59 80.82 59.00 68.21 e-sbar 503 489 451 92.23 89.66 90.93 i-adjp 152 144 115 79.86 75.66 77.70 e-prt 126 129 121 93.80 96.03 94.90 i-sbar 12 9 5 55.56 41.67 47.62 i-conjp 24 20 18 90.00 75.00 81.82 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.05 65.44 68.58 Avg2. 46451 46451 44954 96.78 96.78 96.78 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12281 11635 94.74 95.21 94.98 pp 4633 4675 4547 97.26 98.14 97.70 vp 4768 4776 4508 94.39 94.55 94.47 sbar 503 489 441 90.18 87.67 88.91 adjp 384 375 306 81.60 79.69 80.63 advp 822 779 660 84.72 80.29 82.45 prt 126 129 121 93.80 96.03 94.90 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.67 68.78 70.67 Avg2. 23486 23514 22227 94.53 94.64 94.58 Current max chunk-based F1: 94.58 (iteration 22) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 23 Log-likelihood = -241006.642454 Norm (log-likelihood gradient vector) = 8795.105544 Norm (lambda vector) = 124.550329 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4654 4546 97.68 98.12 97.90 i-np 13660 13629 13287 97.49 97.27 97.38 e-np 12220 12262 11993 97.81 98.14 97.97 o 6349 6370 6164 96.77 97.09 96.93 e-vp 4768 4768 4650 97.53 97.53 97.53 i-vp 2602 2678 2519 94.06 96.81 95.42 e-adjp 384 369 319 86.45 83.07 84.73 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 795 689 86.67 83.82 85.22 i-advp 100 76 62 81.58 62.00 70.45 e-sbar 503 507 462 91.12 91.85 91.49 i-adjp 152 141 115 81.56 75.66 78.50 e-prt 126 130 122 93.85 96.83 95.31 i-sbar 12 8 5 62.50 41.67 50.00 i-conjp 24 20 18 90.00 75.00 81.82 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.61 65.82 69.05 Avg2. 46451 46451 44994 96.86 96.86 96.86 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12262 11634 94.88 95.20 95.04 pp 4633 4654 4540 97.55 97.99 97.77 vp 4768 4768 4517 94.74 94.74 94.74 sbar 503 507 452 89.15 89.86 89.50 adjp 384 369 305 82.66 79.43 81.01 advp 822 795 681 85.66 82.85 84.23 prt 126 130 122 93.85 96.83 95.31 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.85 69.31 71.04 Avg2. 23486 23495 22260 94.74 94.78 94.76 Current max chunk-based F1: 94.76 (iteration 23) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 24 Log-likelihood = -233500.231759 Norm (log-likelihood gradient vector) = 8344.356007 Norm (lambda vector) = 126.444930 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4695 4569 97.32 98.62 97.96 i-np 13660 13432 13179 98.12 96.48 97.29 e-np 12220 12364 12040 97.38 98.53 97.95 o 6349 6447 6191 96.03 97.51 96.76 e-vp 4768 4752 4646 97.77 97.44 97.61 i-vp 2602 2668 2518 94.38 96.77 95.56 e-adjp 384 362 320 88.40 83.33 85.79 i-pp 52 36 34 94.44 65.38 77.27 e-advp 822 828 712 85.99 86.62 86.30 i-advp 100 80 67 83.75 67.00 74.44 e-sbar 503 479 451 94.15 89.66 91.85 i-adjp 152 143 115 80.42 75.66 77.97 e-prt 126 130 122 93.85 96.83 95.31 i-sbar 12 8 5 62.50 41.67 50.00 i-conjp 24 18 18 100.00 75.00 85.71 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 73.22 66.14 69.50 Avg2. 46451 46451 44996 96.87 96.87 96.87 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12364 11683 94.49 95.61 95.05 pp 4633 4695 4563 97.19 98.49 97.83 vp 4768 4752 4517 95.05 94.74 94.89 sbar 503 479 441 92.07 87.67 89.82 adjp 384 362 306 84.53 79.69 82.04 advp 822 828 705 85.14 85.77 85.45 prt 126 130 122 93.85 96.83 95.31 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 9 100.00 56.25 72.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.23 69.50 71.79 Avg2. 23486 23619 22346 94.61 95.15 94.88 Current max chunk-based F1: 94.88 (iteration 24) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 25 Log-likelihood = -228448.838444 Norm (log-likelihood gradient vector) = 20159.608459 Norm (lambda vector) = 130.448064 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4685 4568 97.50 98.60 98.05 i-np 13660 13720 13343 97.25 97.68 97.47 e-np 12220 12221 11986 98.08 98.09 98.08 o 6349 6324 6144 97.15 96.77 96.96 e-vp 4768 4746 4646 97.89 97.44 97.67 i-vp 2602 2665 2517 94.45 96.73 95.58 e-adjp 384 361 319 88.37 83.07 85.64 i-pp 52 37 34 91.89 65.38 76.40 e-advp 822 824 713 86.53 86.74 86.63 i-advp 100 81 69 85.19 69.00 76.24 e-sbar 503 482 454 94.19 90.26 92.18 i-adjp 152 141 113 80.14 74.34 77.13 e-prt 126 128 121 94.53 96.03 95.28 i-sbar 12 9 5 55.56 41.67 47.62 i-conjp 24 18 18 100.00 75.00 85.71 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.94 66.15 69.38 Avg2. 46451 46451 45059 97.00 97.00 97.00 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12221 11642 95.26 95.27 95.27 pp 4633 4685 4562 97.37 98.47 97.92 vp 4768 4746 4516 95.15 94.71 94.93 sbar 503 482 444 92.12 88.27 90.15 adjp 384 361 305 84.49 79.43 81.88 advp 822 824 707 85.80 86.01 85.91 prt 126 128 121 94.53 96.03 95.28 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 9 100.00 56.25 72.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.47 69.44 71.87 Avg2. 23486 23456 22306 95.10 94.98 95.04 Current max chunk-based F1: 95.04 (iteration 25) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 26 Log-likelihood = -219385.458312 Norm (log-likelihood gradient vector) = 7489.861597 Norm (lambda vector) = 132.146347 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4690 4570 97.44 98.64 98.04 i-np 13660 13754 13369 97.20 97.87 97.53 e-np 12220 12193 11979 98.24 98.03 98.14 o 6349 6308 6143 97.38 96.76 97.07 e-vp 4768 4745 4645 97.89 97.42 97.66 i-vp 2602 2670 2520 94.38 96.85 95.60 e-adjp 384 361 319 88.37 83.07 85.64 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 825 711 86.18 86.50 86.34 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 480 452 94.17 89.86 91.96 i-adjp 152 141 113 80.14 74.34 77.13 e-prt 126 128 121 94.53 96.03 95.28 i-sbar 12 9 5 55.56 41.67 47.62 i-conjp 24 18 18 100.00 75.00 85.71 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.76 66.13 69.29 Avg2. 46451 46451 45077 97.04 97.04 97.04 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12193 11636 95.43 95.22 95.33 pp 4633 4690 4563 97.29 98.49 97.89 vp 4768 4745 4516 95.17 94.71 94.94 sbar 503 480 442 92.08 87.87 89.93 adjp 384 361 306 84.76 79.69 82.15 advp 822 825 705 85.45 85.77 85.61 prt 126 128 121 94.53 96.03 95.28 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 9 100.00 56.25 72.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.47 69.40 71.85 Avg2. 23486 23431 22298 95.16 94.94 95.05 Current max chunk-based F1: 95.05 (iteration 26) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 27 Log-likelihood = -216228.128800 Norm (log-likelihood gradient vector) = 8004.229819 Norm (lambda vector) = 132.047529 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4696 4574 97.40 98.73 98.06 i-np 13660 13755 13369 97.19 97.87 97.53 e-np 12220 12183 11973 98.28 97.98 98.13 o 6349 6301 6142 97.48 96.74 97.11 e-vp 4768 4746 4648 97.94 97.48 97.71 i-vp 2602 2671 2518 94.27 96.77 95.51 e-adjp 384 365 322 88.22 83.85 85.98 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 827 709 85.73 86.25 85.99 i-advp 100 89 70 78.65 70.00 74.07 e-sbar 503 480 453 94.38 90.06 92.17 i-adjp 152 142 114 80.28 75.00 77.55 e-prt 126 128 121 94.53 96.03 95.28 i-sbar 12 9 6 66.67 50.00 57.14 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 73.02 65.21 68.90 Avg2. 46451 46451 45074 97.04 97.04 97.04 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12183 11626 95.43 95.14 95.28 pp 4633 4696 4567 97.25 98.58 97.91 vp 4768 4746 4520 95.24 94.80 95.02 sbar 503 480 444 92.50 88.27 90.34 adjp 384 365 307 84.11 79.95 81.98 advp 822 827 701 84.76 85.28 85.02 prt 126 128 121 94.53 96.03 95.28 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.38 68.18 71.15 Avg2. 23486 23432 22293 95.14 94.92 95.03 Current max chunk-based F1: 95.05 (iteration 26) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 28 Log-likelihood = -210437.405170 Norm (log-likelihood gradient vector) = 8484.074154 Norm (lambda vector) = 132.851646 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4697 4575 97.40 98.75 98.07 i-np 13660 13771 13387 97.21 98.00 97.60 e-np 12220 12172 11975 98.38 98.00 98.19 o 6349 6293 6141 97.58 96.72 97.15 e-vp 4768 4750 4649 97.87 97.50 97.69 i-vp 2602 2676 2522 94.25 96.93 95.57 e-adjp 384 376 324 86.17 84.38 85.26 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 814 703 86.36 85.52 85.94 i-advp 100 89 69 77.53 69.00 73.02 e-sbar 503 477 452 94.76 89.86 92.24 i-adjp 152 145 115 79.31 75.66 77.44 e-prt 126 128 121 94.53 96.03 95.28 i-sbar 12 9 6 66.67 50.00 57.14 i-conjp 24 10 10 100.00 41.67 58.82 e-conjp 16 5 5 100.00 31.25 47.62 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 72.76 63.73 67.95 Avg2. 46451 46451 45088 97.07 97.07 97.07 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12173 11641 95.63 95.26 95.45 pp 4633 4697 4568 97.25 98.60 97.92 vp 4768 4750 4526 95.28 94.92 95.10 sbar 503 477 443 92.87 88.07 90.41 adjp 384 376 309 82.18 80.47 81.32 advp 822 814 696 85.50 84.67 85.09 prt 126 128 121 94.53 96.03 95.28 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 5 5 100.00 31.25 47.62 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.33 66.93 70.43 Avg2. 23486 23420 22309 95.26 94.99 95.12 Current max chunk-based F1: 95.12 (iteration 28) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 29 Log-likelihood = -202475.984056 Norm (log-likelihood gradient vector) = 7695.595915 Norm (lambda vector) = 135.355785 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4558 97.64 98.38 98.01 i-np 13660 13623 13308 97.69 97.42 97.56 e-np 12220 12231 12007 98.17 98.26 98.21 o 6349 6367 6172 96.94 97.21 97.07 e-vp 4768 4755 4649 97.77 97.50 97.64 i-vp 2602 2675 2528 94.50 97.16 95.81 e-adjp 384 384 328 85.42 85.42 85.42 i-pp 52 37 34 91.89 65.38 76.40 e-advp 822 837 711 84.95 86.50 85.71 i-advp 100 88 67 76.14 67.00 71.28 e-sbar 503 505 461 91.29 91.65 91.47 i-adjp 152 142 114 80.28 75.00 77.55 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 9 6 66.67 50.00 57.14 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 62.70 60.23 61.44 Avg2. 46451 46451 45066 97.02 97.02 97.02 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12232 11671 95.41 95.51 95.46 pp 4633 4668 4551 97.49 98.23 97.86 vp 4768 4755 4534 95.35 95.09 95.22 sbar 503 505 452 89.50 89.86 89.68 adjp 384 384 312 81.25 81.25 81.25 advp 822 837 702 83.87 85.40 84.63 prt 126 130 123 94.62 97.62 96.09 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.75 64.30 64.02 Avg2. 23486 23511 22345 95.04 95.14 95.09 Current max chunk-based F1: 95.12 (iteration 28) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 30 Log-likelihood = -195332.967742 Norm (log-likelihood gradient vector) = 12193.907892 Norm (lambda vector) = 142.007321 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4698 4575 97.38 98.75 98.06 i-np 13660 13697 13346 97.44 97.70 97.57 e-np 12220 12199 11993 98.31 98.14 98.23 o 6349 6327 6156 97.30 96.96 97.13 e-vp 4768 4754 4649 97.79 97.50 97.65 i-vp 2602 2684 2531 94.30 97.27 95.76 e-adjp 384 381 328 86.09 85.42 85.75 i-pp 52 37 34 91.89 65.38 76.40 e-advp 822 835 707 84.67 86.01 85.33 i-advp 100 85 65 76.47 65.00 70.27 e-sbar 503 480 450 93.75 89.46 91.56 i-adjp 152 138 114 82.61 75.00 78.62 e-prt 126 128 121 94.53 96.03 95.28 i-sbar 12 8 6 75.00 50.00 60.00 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.38 59.93 61.61 Avg2. 46451 46451 45075 97.04 97.04 97.04 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12200 11655 95.53 95.38 95.45 pp 4633 4698 4567 97.21 98.58 97.89 vp 4768 4754 4532 95.33 95.05 95.19 sbar 503 480 443 92.29 88.07 90.13 adjp 384 381 312 81.89 81.25 81.57 advp 822 835 696 83.35 84.67 84.01 prt 126 128 121 94.53 96.03 95.28 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 64.01 63.90 63.96 Avg2. 23486 23476 22326 95.10 95.06 95.08 Current max chunk-based F1: 95.12 (iteration 28) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 31 Log-likelihood = -186766.773053 Norm (log-likelihood gradient vector) = 6260.851158 Norm (lambda vector) = 144.672303 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4690 4569 97.42 98.62 98.02 i-np 13660 13670 13331 97.52 97.59 97.56 e-np 12220 12218 11999 98.21 98.19 98.20 o 6349 6349 6168 97.15 97.15 97.15 e-vp 4768 4757 4649 97.73 97.50 97.62 i-vp 2602 2676 2527 94.43 97.12 95.76 e-adjp 384 376 326 86.70 84.90 85.79 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 835 713 85.39 86.74 86.06 i-advp 100 84 66 78.57 66.00 71.74 e-sbar 503 486 453 93.21 90.06 91.61 i-adjp 152 136 113 83.09 74.34 78.47 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 9 7 77.78 58.33 66.67 i-conjp 24 0 0 0.00 0.00 0.00 e-conjp 16 0 0 0.00 0.00 0.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.95 60.48 62.16 Avg2. 46451 46451 45078 97.04 97.04 97.04 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12219 11665 95.47 95.46 95.46 pp 4633 4690 4562 97.27 98.47 97.87 vp 4768 4757 4532 95.27 95.05 95.16 sbar 503 486 446 91.77 88.67 90.19 adjp 384 376 312 82.98 81.25 82.11 advp 822 835 704 84.31 85.64 84.97 prt 126 130 123 94.62 97.62 96.09 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 64.17 64.22 64.19 Avg2. 23486 23493 22344 95.11 95.14 95.12 Current max chunk-based F1: 95.12 (iteration 31) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 32 Log-likelihood = -182333.073904 Norm (log-likelihood gradient vector) = 5203.232530 Norm (lambda vector) = 144.842518 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4694 4577 97.51 98.79 98.15 i-np 13660 13677 13342 97.55 97.67 97.61 e-np 12220 12228 12007 98.19 98.26 98.22 o 6349 6355 6171 97.10 97.20 97.15 e-vp 4768 4763 4653 97.69 97.59 97.64 i-vp 2602 2664 2523 94.71 96.96 95.82 e-adjp 384 366 323 88.25 84.11 86.13 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 822 712 86.62 86.62 86.62 i-advp 100 79 66 83.54 66.00 73.74 e-sbar 503 482 453 93.98 90.06 91.98 i-adjp 152 135 113 83.70 74.34 78.75 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 9 7 77.78 58.33 66.67 i-conjp 24 8 8 100.00 33.33 50.00 e-conjp 16 4 4 100.00 25.00 40.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.42 63.36 68.45 Avg2. 46451 46451 45116 97.13 97.13 97.13 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12229 11687 95.57 95.64 95.60 pp 4633 4694 4570 97.36 98.64 98.00 vp 4768 4763 4536 95.23 95.13 95.18 sbar 503 482 446 92.53 88.67 90.56 adjp 384 366 310 84.70 80.73 82.67 advp 822 822 704 85.64 85.64 85.64 prt 126 130 123 94.62 97.62 96.09 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 4 4 100.00 25.00 40.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.57 66.71 70.42 Avg2. 23486 23490 22380 95.27 95.29 95.28 Current max chunk-based F1: 95.28 (iteration 32) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 33 Log-likelihood = -174232.240618 Norm (log-likelihood gradient vector) = 5538.925864 Norm (lambda vector) = 147.271557 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4646 4555 98.04 98.32 98.18 i-np 13660 13570 13303 98.03 97.39 97.71 e-np 12220 12303 12045 97.90 98.57 98.23 o 6349 6436 6213 96.54 97.86 97.19 e-vp 4768 4756 4651 97.79 97.55 97.67 i-vp 2602 2660 2521 94.77 96.89 95.82 e-adjp 384 354 319 90.11 83.07 86.45 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 814 710 87.22 86.37 86.80 i-advp 100 77 67 87.01 67.00 75.71 e-sbar 503 512 468 91.41 93.04 92.22 i-adjp 152 127 113 88.98 74.34 81.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 9 6 66.67 50.00 57.14 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.31 65.31 69.52 Avg2. 46451 46451 45150 97.20 97.20 97.20 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12304 11737 95.39 96.05 95.72 pp 4633 4646 4548 97.89 98.17 98.03 vp 4768 4756 4538 95.42 95.18 95.30 sbar 503 512 459 89.65 91.25 90.44 adjp 384 354 307 86.72 79.95 83.20 advp 822 814 705 86.61 85.77 86.19 prt 126 131 124 94.66 98.41 96.50 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.63 68.85 71.63 Avg2. 23486 23524 22425 95.33 95.48 95.41 Current max chunk-based F1: 95.41 (iteration 33) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 34 Log-likelihood = -168170.529054 Norm (log-likelihood gradient vector) = 12435.771794 Norm (lambda vector) = 151.099608 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4567 98.07 98.58 98.32 i-np 13660 13858 13448 97.04 98.45 97.74 e-np 12220 12156 11977 98.53 98.01 98.27 o 6349 6294 6146 97.65 96.80 97.22 e-vp 4768 4757 4651 97.77 97.55 97.66 i-vp 2602 2665 2524 94.71 97.00 95.84 e-adjp 384 354 318 89.83 82.81 86.18 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 806 709 87.97 86.25 87.10 i-advp 100 79 68 86.08 68.00 75.98 e-sbar 503 500 465 93.00 92.45 92.72 i-adjp 152 128 113 88.28 74.34 80.71 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 10 7 70.00 58.33 63.64 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.54 65.72 69.85 Avg2. 46451 46451 45172 97.25 97.25 97.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12157 11670 95.99 95.50 95.75 pp 4633 4657 4560 97.92 98.42 98.17 vp 4768 4757 4539 95.42 95.20 95.31 sbar 503 500 457 91.40 90.85 91.13 adjp 384 354 306 86.44 79.69 82.93 advp 822 806 704 87.34 85.64 86.49 prt 126 131 124 94.66 98.41 96.50 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.92 68.75 71.70 Avg2. 23486 23369 22367 95.71 95.24 95.47 Current max chunk-based F1: 95.47 (iteration 34) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 35 Log-likelihood = -164138.467857 Norm (log-likelihood gradient vector) = 11376.422063 Norm (lambda vector) = 152.057076 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4571 97.94 98.66 98.30 i-np 13660 13748 13395 97.43 98.06 97.75 e-np 12220 12203 12001 98.34 98.21 98.28 o 6349 6341 6171 97.32 97.20 97.26 e-vp 4768 4757 4652 97.79 97.57 97.68 i-vp 2602 2666 2525 94.71 97.04 95.86 e-adjp 384 359 323 89.97 84.11 86.94 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 806 710 88.09 86.37 87.22 i-advp 100 81 68 83.95 68.00 75.14 e-sbar 503 492 460 93.50 91.45 92.46 i-adjp 152 134 114 85.07 75.00 79.72 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 10 7 70.00 58.33 63.64 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.30 65.79 69.79 Avg2. 46451 46451 45176 97.26 97.26 97.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12204 11698 95.85 95.73 95.79 pp 4633 4667 4564 97.79 98.51 98.15 vp 4768 4757 4540 95.44 95.22 95.33 sbar 503 492 452 91.87 89.86 90.85 adjp 384 359 310 86.35 80.73 83.45 advp 822 806 704 87.34 85.64 86.49 prt 126 131 124 94.66 98.41 96.50 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.93 68.79 71.73 Avg2. 23486 23423 22399 95.63 95.37 95.50 Current max chunk-based F1: 95.50 (iteration 35) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 36 Log-likelihood = -162618.126665 Norm (log-likelihood gradient vector) = 5058.760602 Norm (lambda vector) = 151.634185 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4675 4575 97.86 98.75 98.30 i-np 13660 13687 13359 97.60 97.80 97.70 e-np 12220 12227 12009 98.22 98.27 98.25 o 6349 6368 6179 97.03 97.32 97.18 e-vp 4768 4749 4649 97.89 97.50 97.70 i-vp 2602 2671 2524 94.50 97.00 95.73 e-adjp 384 363 324 89.26 84.38 86.75 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 805 709 88.07 86.25 87.15 i-advp 100 84 71 84.52 71.00 77.17 e-sbar 503 488 458 93.85 91.05 92.43 i-adjp 152 136 114 83.82 75.00 79.17 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 10 7 70.00 58.33 63.64 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.19 65.93 69.81 Avg2. 46451 46451 45157 97.21 97.21 97.21 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11701 95.69 95.75 95.72 pp 4633 4675 4568 97.71 98.60 98.15 vp 4768 4749 4534 95.47 95.09 95.28 sbar 503 488 450 92.21 89.46 90.82 adjp 384 363 311 85.67 80.99 83.27 advp 822 805 705 87.58 85.77 86.66 prt 126 132 124 93.94 98.41 96.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.83 68.78 71.68 Avg2. 23486 23447 22400 95.53 95.38 95.46 Current max chunk-based F1: 95.50 (iteration 35) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 37 Log-likelihood = -160625.684578 Norm (log-likelihood gradient vector) = 3908.436913 Norm (lambda vector) = 152.480423 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4677 4576 97.84 98.77 98.30 i-np 13660 13654 13336 97.67 97.63 97.65 e-np 12220 12243 12015 98.14 98.32 98.23 o 6349 6377 6179 96.90 97.32 97.11 e-vp 4768 4744 4647 97.96 97.46 97.71 i-vp 2602 2671 2528 94.65 97.16 95.88 e-adjp 384 370 326 88.11 84.90 86.47 i-pp 52 36 34 94.44 65.38 77.27 e-advp 822 802 707 88.15 86.01 87.07 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 490 459 93.67 91.25 92.45 i-adjp 152 138 114 82.61 75.00 78.62 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 11 8 72.73 66.67 69.57 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.02 66.37 69.98 Avg2. 46451 46451 45145 97.19 97.19 97.19 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12244 11704 95.59 95.78 95.68 pp 4633 4677 4569 97.69 98.62 98.15 vp 4768 4744 4537 95.64 95.16 95.40 sbar 503 490 452 92.24 89.86 91.04 adjp 384 370 312 84.32 81.25 82.76 advp 822 802 703 87.66 85.52 86.58 prt 126 132 124 93.94 98.41 96.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.71 68.83 71.65 Avg2. 23486 23466 22408 95.49 95.41 95.45 Current max chunk-based F1: 95.50 (iteration 35) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 38 Log-likelihood = -158049.202298 Norm (log-likelihood gradient vector) = 4558.962480 Norm (lambda vector) = 153.965721 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4681 4576 97.76 98.77 98.26 i-np 13660 13634 13341 97.85 97.66 97.76 e-np 12220 12255 12030 98.16 98.45 98.30 o 6349 6382 6191 97.01 97.51 97.26 e-vp 4768 4742 4647 98.00 97.46 97.73 i-vp 2602 2662 2523 94.78 96.96 95.86 e-adjp 384 373 324 86.86 84.38 85.60 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 806 706 87.59 85.89 86.73 i-advp 100 85 70 82.35 70.00 75.68 e-sbar 503 488 456 93.44 90.66 92.03 i-adjp 152 140 114 81.43 75.00 78.08 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 73.95 66.87 70.23 Avg2. 46451 46451 45168 97.24 97.24 97.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12255 11730 95.72 95.99 95.85 pp 4633 4681 4569 97.61 98.62 98.11 vp 4768 4742 4543 95.80 95.28 95.54 sbar 503 488 451 92.42 89.66 91.02 adjp 384 373 311 83.38 80.99 82.17 advp 822 806 702 87.10 85.40 86.24 prt 126 132 124 93.94 98.41 96.12 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.60 68.81 71.59 Avg2. 23486 23484 22437 95.54 95.53 95.54 Current max chunk-based F1: 95.54 (iteration 38) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 39 Log-likelihood = -152207.089679 Norm (log-likelihood gradient vector) = 5845.135832 Norm (lambda vector) = 157.468132 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4571 97.88 98.66 98.27 i-np 13660 13682 13310 97.28 97.44 97.36 e-np 12220 12245 11997 97.97 98.18 98.07 o 6349 6351 6148 96.80 96.83 96.82 e-vp 4768 4742 4649 98.04 97.50 97.77 i-vp 2602 2657 2525 95.03 97.04 96.03 e-adjp 384 377 327 86.74 85.16 85.94 i-pp 52 37 36 97.30 69.23 80.90 e-advp 822 808 707 87.50 86.01 86.75 i-advp 100 89 71 79.78 71.00 75.13 e-sbar 503 490 456 93.06 90.66 91.84 i-adjp 152 134 112 83.58 73.68 78.32 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 73.96 67.57 70.62 Avg2. 46451 46451 45066 97.02 97.02 97.02 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12245 11657 95.20 95.39 95.30 pp 4633 4670 4564 97.73 98.51 98.12 vp 4768 4742 4547 95.89 95.36 95.63 sbar 503 490 450 91.84 89.46 90.63 adjp 384 377 313 83.02 81.51 82.26 advp 822 808 700 86.63 85.16 85.89 prt 126 133 124 93.23 98.41 95.75 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.35 69.38 71.78 Avg2. 23486 23473 22363 95.27 95.22 95.24 Current max chunk-based F1: 95.54 (iteration 38) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 40 Log-likelihood = -150194.350002 Norm (log-likelihood gradient vector) = 12386.339012 Norm (lambda vector) = 162.302018 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4671 4571 97.86 98.66 98.26 i-np 13660 13674 13352 97.65 97.75 97.70 e-np 12220 12249 12023 98.15 98.39 98.27 o 6349 6365 6174 97.00 97.24 97.12 e-vp 4768 4739 4647 98.06 97.46 97.76 i-vp 2602 2667 2530 94.86 97.23 96.03 e-adjp 384 372 325 87.37 84.64 85.98 i-pp 52 36 35 97.22 67.31 79.55 e-advp 822 808 710 87.87 86.37 87.12 i-advp 100 86 70 81.40 70.00 75.27 e-sbar 503 485 454 93.61 90.26 91.90 i-adjp 152 130 112 86.15 73.68 79.43 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.27 67.45 70.70 Avg2. 46451 46451 45160 97.22 97.22 97.22 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12249 11711 95.61 95.83 95.72 pp 4633 4671 4564 97.71 98.51 98.11 vp 4768 4739 4544 95.89 95.30 95.59 sbar 503 485 448 92.37 89.07 90.69 adjp 384 372 312 83.87 81.25 82.54 advp 822 808 704 87.13 85.64 86.38 prt 126 133 124 93.23 98.41 95.75 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.58 69.40 71.90 Avg2. 23486 23465 22415 95.53 95.44 95.48 Current max chunk-based F1: 95.54 (iteration 38) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 41 Log-likelihood = -143565.363311 Norm (log-likelihood gradient vector) = 4501.542381 Norm (lambda vector) = 162.359590 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4571 97.88 98.66 98.27 i-np 13660 13710 13390 97.67 98.02 97.84 e-np 12220 12230 12025 98.32 98.40 98.36 o 6349 6355 6184 97.31 97.40 97.36 e-vp 4768 4744 4649 98.00 97.50 97.75 i-vp 2602 2660 2529 95.08 97.19 96.12 e-adjp 384 373 324 86.86 84.38 85.60 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 807 709 87.86 86.25 87.05 i-advp 100 86 70 81.40 70.00 75.27 e-sbar 503 484 454 93.80 90.26 92.00 i-adjp 152 128 110 85.94 72.37 78.57 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.27 67.30 70.61 Avg2. 46451 46451 45206 97.32 97.32 97.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11730 95.91 95.99 95.95 pp 4633 4670 4564 97.73 98.51 98.12 vp 4768 4744 4545 95.81 95.32 95.56 sbar 503 484 448 92.56 89.07 90.78 adjp 384 373 310 83.11 80.73 81.90 advp 822 807 703 87.11 85.52 86.31 prt 126 133 124 93.23 98.41 95.75 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.55 69.36 71.86 Avg2. 23486 23449 22432 95.66 95.51 95.59 Current max chunk-based F1: 95.59 (iteration 41) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 42 Log-likelihood = -141331.767623 Norm (log-likelihood gradient vector) = 3033.500894 Norm (lambda vector) = 162.538495 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4571 97.96 98.66 98.31 i-np 13660 13761 13412 97.46 98.18 97.82 e-np 12220 12205 12010 98.40 98.28 98.34 o 6349 6330 6170 97.47 97.18 97.33 e-vp 4768 4750 4655 98.00 97.63 97.81 i-vp 2602 2663 2531 95.04 97.27 96.14 e-adjp 384 372 325 87.37 84.64 85.98 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 807 709 87.86 86.25 87.05 i-advp 100 85 70 82.35 70.00 75.68 e-sbar 503 486 458 94.24 91.05 92.62 i-adjp 152 122 108 88.52 71.05 78.83 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 0 0 0.00 0.00 0.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.50 67.28 70.71 Avg2. 46451 46451 45210 97.33 97.33 97.33 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12206 11719 96.01 95.90 95.96 pp 4633 4666 4564 97.81 98.51 98.16 vp 4768 4750 4549 95.77 95.41 95.59 sbar 503 486 452 93.00 89.86 91.41 adjp 384 372 310 83.33 80.73 82.01 advp 822 807 703 87.11 85.52 86.31 prt 126 133 124 93.23 98.41 95.75 lst 10 0 0 0.00 0.00 0.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.63 69.43 71.94 Avg2. 23486 23428 22429 95.74 95.50 95.62 Current max chunk-based F1: 95.62 (iteration 42) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 43 Log-likelihood = -138124.348208 Norm (log-likelihood gradient vector) = 3843.398495 Norm (lambda vector) = 163.461714 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4569 98.01 98.62 98.31 i-np 13660 13584 13326 98.10 97.55 97.83 e-np 12220 12291 12051 98.05 98.62 98.33 o 6349 6411 6213 96.91 97.86 97.38 e-vp 4768 4752 4654 97.94 97.61 97.77 i-vp 2602 2670 2532 94.83 97.31 96.05 e-adjp 384 375 325 86.67 84.64 85.64 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 805 708 87.95 86.13 87.03 i-advp 100 80 68 85.00 68.00 75.56 e-sbar 503 490 460 93.88 91.45 92.65 i-adjp 152 122 108 88.52 71.05 78.83 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 3 3 100.00 30.00 46.15 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.71 68.72 73.37 Avg2. 46451 46451 45208 97.32 97.32 97.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12291 11762 95.70 96.25 95.97 pp 4633 4662 4562 97.85 98.47 98.16 vp 4768 4752 4547 95.69 95.36 95.53 sbar 503 490 454 92.65 90.26 91.44 adjp 384 375 310 82.67 80.73 81.69 advp 822 805 703 87.33 85.52 86.42 prt 126 133 124 93.23 98.41 95.75 lst 10 3 3 100.00 30.00 46.15 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.40 72.50 77.57 Avg2. 23486 23520 22473 95.55 95.69 95.62 Current max chunk-based F1: 95.62 (iteration 43) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 44 Log-likelihood = -135662.269582 Norm (log-likelihood gradient vector) = 8410.017489 Norm (lambda vector) = 165.035771 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4656 4568 98.11 98.60 98.35 i-np 13660 13730 13409 97.66 98.16 97.91 e-np 12220 12217 12019 98.38 98.36 98.37 o 6349 6347 6181 97.38 97.35 97.37 e-vp 4768 4748 4655 98.04 97.63 97.84 i-vp 2602 2668 2533 94.94 97.35 96.13 e-adjp 384 376 329 87.50 85.68 86.58 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 806 706 87.59 85.89 86.73 i-advp 100 80 67 83.75 67.00 74.44 e-sbar 503 493 460 93.31 91.45 92.37 i-adjp 152 120 107 89.17 70.39 78.68 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 4 4 100.00 40.00 57.14 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.71 69.17 73.63 Avg2. 46451 46451 45229 97.37 97.37 97.37 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12218 11736 96.06 96.04 96.05 pp 4633 4656 4561 97.96 98.45 98.20 vp 4768 4748 4547 95.77 95.36 95.57 sbar 503 493 454 92.09 90.26 91.16 adjp 384 376 313 83.24 81.51 82.37 advp 822 806 699 86.72 85.04 85.87 prt 126 133 124 93.23 98.41 95.75 lst 10 4 4 100.00 40.00 57.14 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.40 73.51 78.14 Avg2. 23486 23443 22446 95.75 95.57 95.66 Current max chunk-based F1: 95.66 (iteration 44) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 45 Log-likelihood = -132137.011542 Norm (log-likelihood gradient vector) = 4847.524336 Norm (lambda vector) = 166.766037 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4571 98.05 98.66 98.35 i-np 13660 13734 13412 97.66 98.18 97.92 e-np 12220 12213 12018 98.40 98.35 98.38 o 6349 6344 6180 97.41 97.34 97.38 e-vp 4768 4749 4656 98.04 97.65 97.85 i-vp 2602 2667 2533 94.98 97.35 96.15 e-adjp 384 376 329 87.50 85.68 86.58 i-pp 52 35 34 97.14 65.38 78.16 e-advp 822 805 706 87.70 85.89 86.79 i-advp 100 81 68 83.95 68.00 75.14 e-sbar 503 490 458 93.47 91.05 92.25 i-adjp 152 120 107 89.17 70.39 78.68 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 4 4 100.00 40.00 57.14 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.74 69.20 73.66 Avg2. 46451 46451 45233 97.38 97.38 97.38 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12214 11731 96.05 96.00 96.02 pp 4633 4662 4564 97.90 98.51 98.20 vp 4768 4749 4548 95.77 95.39 95.58 sbar 503 490 452 92.24 89.86 91.04 adjp 384 376 313 83.24 81.51 82.37 advp 822 805 700 86.96 85.16 86.05 prt 126 133 124 93.23 98.41 95.75 lst 10 4 4 100.00 40.00 57.14 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.43 73.48 78.14 Avg2. 23486 23442 22444 95.74 95.56 95.65 Current max chunk-based F1: 95.66 (iteration 44) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 46 Log-likelihood = -130679.041166 Norm (log-likelihood gradient vector) = 4019.648957 Norm (lambda vector) = 167.209962 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4577 98.09 98.79 98.44 i-np 13660 13682 13379 97.79 97.94 97.86 e-np 12220 12237 12023 98.25 98.39 98.32 o 6349 6354 6182 97.29 97.37 97.33 e-vp 4768 4745 4657 98.15 97.67 97.91 i-vp 2602 2668 2539 95.16 97.58 96.36 e-adjp 384 372 328 88.17 85.42 86.77 i-pp 52 37 36 97.30 69.23 80.90 e-advp 822 815 713 87.48 86.74 87.11 i-advp 100 81 68 83.95 68.00 75.14 e-sbar 503 493 461 93.51 91.65 92.57 i-adjp 152 126 111 88.10 73.03 79.86 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 4 4 100.00 40.00 57.14 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.72 69.59 73.88 Avg2. 46451 46451 45235 97.38 97.38 97.38 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12237 11731 95.86 96.00 95.93 pp 4633 4666 4570 97.94 98.64 98.29 vp 4768 4745 4555 96.00 95.53 95.76 sbar 503 493 455 92.29 90.46 91.37 adjp 384 372 313 84.14 81.51 82.80 advp 822 815 705 86.50 85.77 86.13 prt 126 133 124 93.23 98.41 95.75 lst 10 4 4 100.00 40.00 57.14 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.49 73.63 78.25 Avg2. 23486 23474 22465 95.70 95.65 95.68 Current max chunk-based F1: 95.68 (iteration 46) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 47 Log-likelihood = -127309.540869 Norm (log-likelihood gradient vector) = 3474.200319 Norm (lambda vector) = 168.809769 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4679 4584 97.97 98.94 98.45 i-np 13660 13668 13368 97.81 97.86 97.83 e-np 12220 12248 12026 98.19 98.41 98.30 o 6349 6352 6177 97.24 97.29 97.27 e-vp 4768 4743 4658 98.21 97.69 97.95 i-vp 2602 2672 2539 95.02 97.58 96.28 e-adjp 384 370 329 88.92 85.68 87.27 i-pp 52 40 36 90.00 69.23 78.26 e-advp 822 810 709 87.53 86.25 86.89 i-advp 100 81 68 83.95 68.00 75.14 e-sbar 503 481 455 94.59 90.46 92.48 i-adjp 152 132 112 84.85 73.68 78.87 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 4 4 100.00 40.00 57.14 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.28 69.56 73.66 Avg2. 46451 46451 45222 97.35 97.35 97.35 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12248 11735 95.81 96.03 95.92 pp 4633 4679 4575 97.78 98.75 98.26 vp 4768 4743 4551 95.95 95.45 95.70 sbar 503 481 450 93.56 89.46 91.46 adjp 384 370 312 84.32 81.25 82.76 advp 822 810 701 86.54 85.28 85.91 prt 126 133 124 93.23 98.41 95.75 lst 10 4 4 100.00 40.00 57.14 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.61 73.46 78.21 Avg2. 23486 23477 22460 95.67 95.63 95.65 Current max chunk-based F1: 95.68 (iteration 46) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 48 Log-likelihood = -124273.675393 Norm (log-likelihood gradient vector) = 3991.430429 Norm (lambda vector) = 170.455838 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4577 98.05 98.79 98.42 i-np 13660 13644 13359 97.91 97.80 97.85 e-np 12220 12256 12036 98.20 98.49 98.35 o 6349 6351 6176 97.24 97.28 97.26 e-vp 4768 4741 4654 98.16 97.61 97.89 i-vp 2602 2686 2539 94.53 97.58 96.03 e-adjp 384 371 330 88.95 85.94 87.42 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 808 709 87.75 86.25 86.99 i-advp 100 82 68 82.93 68.00 74.73 e-sbar 503 489 459 93.87 91.25 92.54 i-adjp 152 139 114 82.01 75.00 78.35 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.97 70.16 73.86 Avg2. 46451 46451 45219 97.35 97.35 97.35 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12256 11747 95.85 96.13 95.99 pp 4633 4668 4567 97.84 98.58 98.20 vp 4768 4741 4542 95.80 95.26 95.53 sbar 503 489 453 92.64 90.06 91.33 adjp 384 371 313 84.37 81.51 82.91 advp 822 808 701 86.76 85.28 86.01 prt 126 132 124 93.94 98.41 96.12 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.61 74.52 78.80 Avg2. 23486 23479 22460 95.66 95.63 95.65 Current max chunk-based F1: 95.68 (iteration 46) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 49 Log-likelihood = -120337.796023 Norm (log-likelihood gradient vector) = 4067.096919 Norm (lambda vector) = 173.373834 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4690 4586 97.78 98.99 98.38 i-np 13660 13716 13403 97.72 98.12 97.92 e-np 12220 12230 12030 98.36 98.45 98.40 o 6349 6310 6160 97.62 97.02 97.32 e-vp 4768 4737 4652 98.21 97.57 97.89 i-vp 2602 2683 2538 94.60 97.54 96.05 e-adjp 384 370 329 88.92 85.68 87.27 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 809 710 87.76 86.37 87.06 i-advp 100 83 68 81.93 68.00 74.32 e-sbar 503 475 451 94.95 89.66 92.23 i-adjp 152 135 113 83.70 74.34 78.75 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 15 14 93.33 58.33 71.79 e-conjp 16 8 7 87.50 43.75 58.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.96 69.32 73.39 Avg2. 46451 46451 45235 97.38 97.38 97.38 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11745 96.03 96.11 96.07 pp 4633 4690 4576 97.57 98.77 98.17 vp 4768 4737 4542 95.88 95.26 95.57 sbar 503 475 446 93.89 88.67 91.21 adjp 384 370 313 84.59 81.51 83.02 advp 822 809 702 86.77 85.40 86.08 prt 126 132 124 93.94 98.41 96.12 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 7 87.50 43.75 58.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.62 73.79 78.40 Avg2. 23486 23456 22460 95.75 95.63 95.69 Current max chunk-based F1: 95.69 (iteration 49) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 50 Log-likelihood = -117336.677246 Norm (log-likelihood gradient vector) = 4752.793616 Norm (lambda vector) = 175.769138 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4681 4580 97.84 98.86 98.35 i-np 13660 13652 13382 98.02 97.96 97.99 e-np 12220 12258 12048 98.29 98.59 98.44 o 6349 6359 6188 97.31 97.46 97.39 e-vp 4768 4742 4652 98.10 97.57 97.83 i-vp 2602 2679 2533 94.55 97.35 95.93 e-adjp 384 370 329 88.92 85.68 87.27 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 805 708 87.95 86.13 87.03 i-advp 100 82 68 82.93 68.00 74.73 e-sbar 503 478 450 94.14 89.46 91.74 i-adjp 152 133 112 84.21 73.68 78.60 e-prt 126 134 124 92.54 98.41 95.38 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 15 14 93.33 58.33 71.79 e-conjp 16 8 7 87.50 43.75 58.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.01 69.08 73.27 Avg2. 46451 46451 45243 97.40 97.40 97.40 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12258 11774 96.05 96.35 96.20 pp 4633 4681 4570 97.63 98.64 98.13 vp 4768 4742 4540 95.74 95.22 95.48 sbar 503 478 445 93.10 88.47 90.72 adjp 384 370 313 84.59 81.51 83.02 advp 822 805 701 87.08 85.28 86.17 prt 126 134 124 92.54 98.41 95.38 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 7 87.50 43.75 58.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.42 73.76 78.30 Avg2. 23486 23481 22479 95.73 95.71 95.72 Current max chunk-based F1: 95.72 (iteration 50) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 51 Log-likelihood = -114832.178663 Norm (log-likelihood gradient vector) = 3173.083131 Norm (lambda vector) = 176.442244 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4570 97.98 98.64 98.31 i-np 13660 13711 13413 97.83 98.19 98.01 e-np 12220 12233 12041 98.43 98.54 98.48 o 6349 6347 6189 97.51 97.48 97.50 e-vp 4768 4753 4656 97.96 97.65 97.80 i-vp 2602 2664 2530 94.97 97.23 96.09 e-adjp 384 370 329 88.92 85.68 87.27 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 800 708 88.50 86.13 87.30 i-advp 100 82 68 82.93 68.00 74.73 e-sbar 503 487 457 93.84 90.85 92.32 i-adjp 152 128 113 88.28 74.34 80.71 e-prt 126 134 124 92.54 98.41 95.38 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 15 14 93.33 58.33 71.79 e-conjp 16 8 7 87.50 43.75 58.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.25 69.18 73.44 Avg2. 46451 46451 45267 97.45 97.45 97.45 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12233 11772 96.23 96.33 96.28 pp 4633 4664 4560 97.77 98.42 98.10 vp 4768 4753 4548 95.69 95.39 95.54 sbar 503 487 451 92.61 89.66 91.11 adjp 384 370 315 85.14 82.03 83.55 advp 822 800 701 87.62 85.28 86.44 prt 126 134 124 92.54 98.41 95.38 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 7 87.50 43.75 58.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.51 73.93 78.43 Avg2. 23486 23454 22483 95.86 95.73 95.79 Current max chunk-based F1: 95.79 (iteration 51) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 52 Log-likelihood = -111938.957539 Norm (log-likelihood gradient vector) = 2948.167010 Norm (lambda vector) = 177.559869 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4570 98.09 98.64 98.36 i-np 13660 13577 13345 98.29 97.69 97.99 e-np 12220 12291 12066 98.17 98.74 98.45 o 6349 6413 6218 96.96 97.94 97.45 e-vp 4768 4756 4657 97.92 97.67 97.80 i-vp 2602 2665 2531 94.97 97.27 96.11 e-adjp 384 375 331 88.27 86.20 87.22 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 804 710 88.31 86.37 87.33 i-advp 100 80 68 85.00 68.00 75.56 e-sbar 503 489 460 94.07 91.45 92.74 i-adjp 152 127 111 87.40 73.03 79.57 e-prt 126 134 124 92.54 98.41 95.38 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.37 69.92 73.91 Avg2. 46451 46451 45263 97.44 97.44 97.44 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12291 11793 95.95 96.51 96.23 pp 4633 4659 4560 97.88 98.42 98.15 vp 4768 4756 4549 95.65 95.41 95.53 sbar 503 489 454 92.84 90.26 91.53 adjp 384 375 316 84.27 82.29 83.27 advp 822 804 704 87.56 85.64 86.59 prt 126 134 124 92.54 98.41 95.38 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.56 74.69 78.88 Avg2. 23486 23522 22513 95.71 95.86 95.78 Current max chunk-based F1: 95.79 (iteration 51) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 53 Log-likelihood = -109957.293435 Norm (log-likelihood gradient vector) = 5222.651535 Norm (lambda vector) = 178.978882 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4569 98.05 98.62 98.33 i-np 13660 13752 13423 97.61 98.27 97.94 e-np 12220 12216 12026 98.44 98.41 98.43 o 6349 6333 6173 97.47 97.23 97.35 e-vp 4768 4755 4660 98.00 97.73 97.87 i-vp 2602 2655 2529 95.25 97.19 96.21 e-adjp 384 367 326 88.83 84.90 86.82 i-pp 52 42 35 83.33 67.31 74.47 e-advp 822 798 707 88.60 86.01 87.28 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 485 458 94.43 91.05 92.71 i-adjp 152 127 110 86.61 72.37 78.85 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.04 69.96 73.78 Avg2. 46451 46451 45248 97.41 97.41 97.41 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12216 11750 96.19 96.15 96.17 pp 4633 4660 4557 97.79 98.36 98.07 vp 4768 4755 4552 95.73 95.47 95.60 sbar 503 485 453 93.40 90.06 91.70 adjp 384 367 312 85.01 81.25 83.09 advp 822 798 700 87.72 85.16 86.42 prt 126 135 124 91.85 98.41 95.02 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.66 74.49 78.81 Avg2. 23486 23430 22461 95.86 95.64 95.75 Current max chunk-based F1: 95.79 (iteration 51) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 54 Log-likelihood = -107644.184265 Norm (log-likelihood gradient vector) = 7046.703267 Norm (lambda vector) = 181.846166 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4571 98.03 98.66 98.34 i-np 13660 13669 13392 97.97 98.04 98.01 e-np 12220 12253 12049 98.34 98.60 98.47 o 6349 6368 6195 97.28 97.57 97.43 e-vp 4768 4752 4658 98.02 97.69 97.86 i-vp 2602 2660 2531 95.15 97.27 96.20 e-adjp 384 370 327 88.38 85.16 86.74 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 800 708 88.50 86.13 87.30 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 487 458 94.05 91.05 92.53 i-adjp 152 128 110 85.94 72.37 78.57 e-prt 126 134 124 92.54 98.41 95.38 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.09 70.00 73.82 Avg2. 46451 46451 45266 97.45 97.45 97.45 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12253 11781 96.15 96.41 96.28 pp 4633 4663 4559 97.77 98.40 98.09 vp 4768 4752 4550 95.75 95.43 95.59 sbar 503 487 453 93.02 90.06 91.52 adjp 384 370 312 84.32 81.25 82.76 advp 822 800 701 87.62 85.28 86.44 prt 126 134 124 92.54 98.41 95.38 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.61 74.52 78.80 Avg2. 23486 23473 22493 95.82 95.77 95.80 Current max chunk-based F1: 95.80 (iteration 54) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 55 Log-likelihood = -106529.468019 Norm (log-likelihood gradient vector) = 3106.212436 Norm (lambda vector) = 181.675633 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4573 97.96 98.70 98.33 i-np 13660 13640 13387 98.15 98.00 98.07 e-np 12220 12263 12056 98.31 98.66 98.48 o 6349 6378 6204 97.27 97.72 97.49 e-vp 4768 4747 4659 98.15 97.71 97.93 i-vp 2602 2659 2531 95.19 97.27 96.22 e-adjp 384 370 328 88.65 85.42 87.00 i-pp 52 40 36 90.00 69.23 78.26 e-advp 822 808 711 88.00 86.50 87.24 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 488 457 93.65 90.85 92.23 i-adjp 152 130 112 86.15 73.68 79.43 e-prt 126 134 124 92.54 98.41 95.38 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.32 70.19 74.03 Avg2. 46451 46451 45286 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12263 11794 96.18 96.51 96.34 pp 4633 4668 4563 97.75 98.49 98.12 vp 4768 4747 4549 95.83 95.41 95.62 sbar 503 488 452 92.62 89.86 91.22 adjp 384 370 313 84.59 81.51 83.02 advp 822 808 704 87.13 85.64 86.38 prt 126 134 124 92.54 98.41 95.38 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.55 74.58 78.81 Avg2. 23486 23492 22512 95.83 95.85 95.84 Current max chunk-based F1: 95.84 (iteration 55) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 56 Log-likelihood = -105503.612888 Norm (log-likelihood gradient vector) = 3432.261934 Norm (lambda vector) = 182.220857 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4576 97.99 98.77 98.38 i-np 13660 13658 13383 97.99 97.97 97.98 e-np 12220 12254 12045 98.29 98.57 98.43 o 6349 6366 6191 97.25 97.51 97.38 e-vp 4768 4740 4656 98.23 97.65 97.94 i-vp 2602 2659 2529 95.11 97.19 96.14 e-adjp 384 369 328 88.89 85.42 87.12 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 809 712 88.01 86.62 87.31 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 488 458 93.85 91.05 92.43 i-adjp 152 137 115 83.94 75.66 79.58 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 5 5 100.00 50.00 66.67 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.07 70.24 73.95 Avg2. 46451 46451 45260 97.44 97.44 97.44 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12254 11775 96.09 96.36 96.22 pp 4633 4670 4566 97.77 98.55 98.16 vp 4768 4740 4545 95.89 95.32 95.60 sbar 503 488 453 92.83 90.06 91.42 adjp 384 369 312 84.55 81.25 82.87 advp 822 809 705 87.14 85.77 86.45 prt 126 135 124 91.85 98.41 95.02 lst 10 5 5 100.00 50.00 66.67 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.50 74.57 78.78 Avg2. 23486 23479 22493 95.80 95.77 95.79 Current max chunk-based F1: 95.84 (iteration 55) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 57 Log-likelihood = -104105.666417 Norm (log-likelihood gradient vector) = 3829.761996 Norm (lambda vector) = 183.063040 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4675 4578 97.93 98.81 98.37 i-np 13660 13651 13374 97.97 97.91 97.94 e-np 12220 12256 12047 98.29 98.58 98.44 o 6349 6356 6186 97.33 97.43 97.38 e-vp 4768 4732 4655 98.37 97.63 98.00 i-vp 2602 2668 2535 95.01 97.43 96.20 e-adjp 384 371 330 88.95 85.94 87.42 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 807 711 88.10 86.50 87.29 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 485 458 94.43 91.05 92.71 i-adjp 152 142 116 81.69 76.32 78.91 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 18 17 94.44 70.83 80.95 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 8 7 87.50 70.00 77.78 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.14 72.14 74.55 Avg2. 46451 46451 45261 97.44 97.44 97.44 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12256 11774 96.07 96.35 96.21 pp 4633 4675 4567 97.69 98.58 98.13 vp 4768 4732 4544 96.03 95.30 95.66 sbar 503 485 453 93.40 90.06 91.70 adjp 384 371 312 84.10 81.25 82.65 advp 822 807 704 87.24 85.64 86.43 prt 126 135 124 91.85 98.41 95.02 lst 10 8 7 87.50 70.00 77.78 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.39 77.18 79.70 Avg2. 23486 23479 22494 95.80 95.78 95.79 Current max chunk-based F1: 95.84 (iteration 55) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 58 Log-likelihood = -100321.892932 Norm (log-likelihood gradient vector) = 4881.128252 Norm (lambda vector) = 186.155409 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4576 97.99 98.77 98.38 i-np 13660 13799 13406 97.15 98.14 97.64 e-np 12220 12185 11993 98.42 98.14 98.28 o 6349 6280 6133 97.66 96.60 97.13 e-vp 4768 4728 4655 98.46 97.63 98.04 i-vp 2602 2675 2542 95.03 97.69 96.34 e-adjp 384 370 329 88.92 85.68 87.27 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 805 714 88.70 86.86 87.77 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 485 457 94.23 90.85 92.51 i-adjp 152 145 117 80.69 76.97 78.79 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 18 17 94.44 70.83 80.95 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.83 72.62 74.67 Avg2. 46451 46451 45194 97.29 97.29 97.29 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12185 11694 95.97 95.70 95.83 pp 4633 4670 4565 97.75 98.53 98.14 vp 4768 4728 4549 96.21 95.41 95.81 sbar 503 485 452 93.20 89.86 91.50 adjp 384 370 312 84.32 81.25 82.76 advp 822 805 705 87.58 85.77 86.66 prt 126 135 124 91.85 98.41 95.02 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.69 78.12 79.86 Avg2. 23486 23398 22418 95.81 95.45 95.63 Current max chunk-based F1: 95.84 (iteration 55) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 59 Log-likelihood = -99019.984701 Norm (log-likelihood gradient vector) = 8739.023251 Norm (lambda vector) = 189.456411 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4673 4578 97.97 98.81 98.39 i-np 13660 13704 13408 97.84 98.16 98.00 e-np 12220 12230 12038 98.43 98.51 98.47 o 6349 6332 6179 97.58 97.32 97.45 e-vp 4768 4730 4656 98.44 97.65 98.04 i-vp 2602 2672 2539 95.02 97.58 96.28 e-adjp 384 370 329 88.92 85.68 87.27 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 806 713 88.46 86.74 87.59 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 485 458 94.43 91.05 92.71 i-adjp 152 140 116 82.86 76.32 79.45 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 18 17 94.44 70.83 80.95 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 8 7 87.50 70.00 77.78 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.30 72.20 74.66 Avg2. 46451 46451 45286 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11770 96.24 96.32 96.28 pp 4633 4673 4567 97.73 98.58 98.15 vp 4768 4730 4549 96.17 95.41 95.79 sbar 503 485 453 93.40 90.06 91.70 adjp 384 370 312 84.32 81.25 82.76 advp 822 806 705 87.47 85.77 86.61 prt 126 135 124 91.85 98.41 95.02 lst 10 8 7 87.50 70.00 77.78 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.47 77.20 79.75 Avg2. 23486 23447 22496 95.94 95.78 95.86 Current max chunk-based F1: 95.86 (iteration 59) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 60 Log-likelihood = -98313.580251 Norm (log-likelihood gradient vector) = 4736.962524 Norm (lambda vector) = 187.805435 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4671 4576 97.97 98.77 98.37 i-np 13660 13705 13415 97.88 98.21 98.04 e-np 12220 12232 12040 98.43 98.53 98.48 o 6349 6337 6185 97.60 97.42 97.51 e-vp 4768 4732 4658 98.44 97.69 98.06 i-vp 2602 2673 2543 95.14 97.73 96.42 e-adjp 384 369 329 89.16 85.68 87.38 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 801 712 88.89 86.62 87.74 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 482 455 94.40 90.46 92.39 i-adjp 152 140 116 82.86 76.32 79.45 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 18 17 94.44 70.83 80.95 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.00 72.58 74.73 Avg2. 46451 46451 45301 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12232 11780 96.30 96.40 96.35 pp 4633 4671 4565 97.73 98.53 98.13 vp 4768 4732 4554 96.24 95.51 95.87 sbar 503 482 450 93.36 89.46 91.37 adjp 384 369 312 84.55 81.25 82.87 advp 822 801 704 87.89 85.64 86.75 prt 126 135 124 91.85 98.41 95.02 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.79 78.15 79.93 Avg2. 23486 23442 22506 96.01 95.83 95.92 Current max chunk-based F1: 95.92 (iteration 60) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 61 Log-likelihood = -95935.359294 Norm (log-likelihood gradient vector) = 2639.905049 Norm (lambda vector) = 189.870487 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4672 4574 97.90 98.73 98.31 i-np 13660 13769 13448 97.67 98.45 98.06 e-np 12220 12199 12027 98.59 98.42 98.51 o 6349 6309 6176 97.89 97.28 97.58 e-vp 4768 4732 4657 98.42 97.67 98.04 i-vp 2602 2678 2545 95.03 97.81 96.40 e-adjp 384 372 330 88.71 85.94 87.30 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 797 713 89.46 86.74 88.08 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 478 451 94.35 89.66 91.95 i-adjp 152 136 116 85.29 76.32 80.56 e-prt 126 135 124 91.85 98.41 95.02 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 18 17 94.44 70.83 80.95 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.13 72.56 74.77 Avg2. 46451 46451 45309 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12199 11769 96.48 96.31 96.39 pp 4633 4672 4563 97.67 98.49 98.08 vp 4768 4732 4553 96.22 95.49 95.85 sbar 503 478 446 93.31 88.67 90.93 adjp 384 372 317 85.22 82.55 83.86 advp 822 797 705 88.46 85.77 87.09 prt 126 135 124 91.85 98.41 95.02 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.92 78.19 80.01 Avg2. 23486 23405 22494 96.11 95.78 95.94 Current max chunk-based F1: 95.94 (iteration 61) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 62 Log-likelihood = -94069.937867 Norm (log-likelihood gradient vector) = 3134.260235 Norm (lambda vector) = 191.166631 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4575 98.01 98.75 98.38 i-np 13660 13700 13413 97.91 98.19 98.05 e-np 12220 12231 12042 98.45 98.54 98.50 o 6349 6344 6192 97.60 97.53 97.57 e-vp 4768 4738 4658 98.31 97.69 98.00 i-vp 2602 2678 2541 94.88 97.66 96.25 e-adjp 384 374 330 88.24 85.94 87.07 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 793 710 89.53 86.37 87.93 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 483 455 94.20 90.46 92.29 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 134 123 91.79 97.62 94.62 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 19 18 94.74 75.00 83.72 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.24 73.09 75.11 Avg2. 46451 46451 45303 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12231 11786 96.36 96.45 96.41 pp 4633 4668 4564 97.77 98.51 98.14 vp 4768 4738 4551 96.05 95.45 95.75 sbar 503 483 450 93.17 89.46 91.28 adjp 384 374 319 85.29 83.07 84.17 advp 822 793 702 88.52 85.40 86.93 prt 126 134 123 91.79 97.62 94.62 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.90 78.22 80.02 Avg2. 23486 23441 22512 96.04 95.85 95.94 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 63 Log-likelihood = -92548.016640 Norm (log-likelihood gradient vector) = 2754.346907 Norm (lambda vector) = 192.481914 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4573 98.07 98.70 98.39 i-np 13660 13970 13484 96.52 98.71 97.60 e-np 12220 12094 11942 98.74 97.73 98.23 o 6349 6214 6097 98.12 96.03 97.06 e-vp 4768 4745 4661 98.23 97.76 97.99 i-vp 2602 2667 2534 95.01 97.39 96.19 e-adjp 384 376 331 88.03 86.20 87.11 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 795 708 89.06 86.13 87.57 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 491 461 93.89 91.65 92.76 i-adjp 152 131 114 87.02 75.00 80.57 e-prt 126 134 123 91.79 97.62 94.62 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.22 72.28 74.67 Avg2. 46451 46451 45174 97.25 97.25 97.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12094 11630 96.16 95.17 95.67 pp 4633 4663 4563 97.86 98.49 98.17 vp 4768 4745 4550 95.89 95.43 95.66 sbar 503 491 456 92.87 90.66 91.75 adjp 384 376 319 84.84 83.07 83.95 advp 822 795 701 88.18 85.28 86.70 prt 126 134 123 91.79 97.62 94.62 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.65 77.57 79.56 Avg2. 23486 23317 22358 95.89 95.20 95.54 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 64 Log-likelihood = -91965.822414 Norm (log-likelihood gradient vector) = 12376.119097 Norm (lambda vector) = 194.580052 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4574 98.07 98.73 98.40 i-np 13660 13845 13462 97.23 98.55 97.89 e-np 12220 12159 11996 98.66 98.17 98.41 o 6349 6277 6149 97.96 96.85 97.40 e-vp 4768 4742 4660 98.27 97.73 98.00 i-vp 2602 2669 2537 95.05 97.50 96.26 e-adjp 384 376 331 88.03 86.20 87.11 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 793 708 89.28 86.13 87.68 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 487 459 94.25 91.25 92.73 i-adjp 152 131 114 87.02 75.00 80.57 e-prt 126 134 123 91.79 97.62 94.62 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 19 18 94.74 75.00 83.72 e-conjp 16 10 9 90.00 56.25 69.23 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.37 73.05 75.15 Avg2. 46451 46451 45262 97.44 97.44 97.44 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12159 11711 96.32 95.83 96.07 pp 4633 4664 4564 97.86 98.51 98.18 vp 4768 4742 4552 95.99 95.47 95.73 sbar 503 487 454 93.22 90.26 91.72 adjp 384 376 319 84.84 83.07 83.95 advp 822 793 700 88.27 85.16 86.69 prt 126 134 123 91.79 97.62 94.62 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 9 90.00 56.25 69.23 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.83 78.22 79.98 Avg2. 23486 23375 22440 96.00 95.55 95.77 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 65 Log-likelihood = -91477.465970 Norm (log-likelihood gradient vector) = 7057.712883 Norm (lambda vector) = 193.570839 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4573 98.05 98.70 98.38 i-np 13660 13709 13412 97.83 98.18 98.01 e-np 12220 12220 12031 98.45 98.45 98.45 o 6349 6344 6187 97.53 97.45 97.49 e-vp 4768 4745 4659 98.19 97.71 97.95 i-vp 2602 2669 2535 94.98 97.43 96.19 e-adjp 384 377 332 88.06 86.46 87.25 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 795 708 89.06 86.13 87.57 i-advp 100 82 69 84.15 69.00 75.82 e-sbar 503 491 461 93.89 91.65 92.76 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 134 123 91.79 97.62 94.62 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 17 16 94.12 66.67 78.05 e-conjp 16 9 8 88.89 50.00 64.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.25 72.41 74.75 Avg2. 46451 46451 45282 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12220 11761 96.24 96.24 96.24 pp 4633 4664 4563 97.83 98.49 98.16 vp 4768 4745 4549 95.87 95.41 95.64 sbar 503 491 456 92.87 90.66 91.75 adjp 384 377 320 84.88 83.33 84.10 advp 822 795 701 88.18 85.28 86.70 prt 126 134 123 91.79 97.62 94.62 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 8 88.89 50.00 64.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.66 77.70 79.63 Avg2. 23486 23445 22489 95.92 95.75 95.84 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 66 Log-likelihood = -90158.603102 Norm (log-likelihood gradient vector) = 3339.952256 Norm (lambda vector) = 194.739913 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4573 98.07 98.70 98.39 i-np 13660 13671 13392 97.96 98.04 98.00 e-np 12220 12234 12037 98.39 98.50 98.45 o 6349 6362 6193 97.34 97.54 97.44 e-vp 4768 4747 4660 98.17 97.73 97.95 i-vp 2602 2671 2535 94.91 97.43 96.15 e-adjp 384 377 331 87.80 86.20 86.99 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 799 710 88.86 86.37 87.60 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 492 461 93.70 91.65 92.66 i-adjp 152 131 115 87.79 75.66 81.27 e-prt 126 134 123 91.79 97.62 94.62 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 15 14 93.33 58.33 71.79 e-conjp 16 8 7 87.50 43.75 58.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.09 71.73 74.31 Avg2. 46451 46451 45274 97.47 97.47 97.47 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12234 11770 96.21 96.32 96.26 pp 4633 4663 4563 97.86 98.49 98.17 vp 4768 4747 4547 95.79 95.36 95.58 sbar 503 492 456 92.68 90.66 91.66 adjp 384 377 320 84.88 83.33 84.10 advp 822 799 703 87.98 85.52 86.74 prt 126 134 123 91.79 97.62 94.62 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 7 87.50 43.75 58.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.47 77.11 79.23 Avg2. 23486 23464 22497 95.88 95.79 95.83 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 67 Log-likelihood = -89295.998777 Norm (log-likelihood gradient vector) = 2493.921805 Norm (lambda vector) = 195.662976 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4574 98.13 98.73 98.43 i-np 13660 13651 13382 98.03 97.96 98.00 e-np 12220 12245 12042 98.34 98.54 98.44 o 6349 6371 6196 97.25 97.59 97.42 e-vp 4768 4742 4660 98.27 97.73 98.00 i-vp 2602 2673 2540 95.02 97.62 96.30 e-adjp 384 376 332 88.30 86.46 87.37 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 807 715 88.60 86.98 87.78 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 492 462 93.90 91.85 92.86 i-adjp 152 130 115 88.46 75.66 81.56 e-prt 126 134 123 91.79 97.62 94.62 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.12 71.84 74.85 Avg2. 46451 46451 45286 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12245 11779 96.19 96.39 96.29 pp 4633 4661 4564 97.92 98.51 98.21 vp 4768 4742 4550 95.95 95.43 95.69 sbar 503 492 457 92.89 90.85 91.86 adjp 384 376 321 85.37 83.59 84.47 advp 822 807 708 87.73 86.13 86.92 prt 126 134 123 91.79 97.62 94.62 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.78 77.23 79.91 Avg2. 23486 23474 22517 95.92 95.87 95.90 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 68 Log-likelihood = -88529.517889 Norm (log-likelihood gradient vector) = 2963.402061 Norm (lambda vector) = 196.296819 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4574 98.07 98.73 98.40 i-np 13660 13659 13387 98.01 98.00 98.01 e-np 12220 12249 12042 98.31 98.54 98.43 o 6349 6366 6195 97.31 97.57 97.44 e-vp 4768 4735 4661 98.44 97.76 98.10 i-vp 2602 2671 2540 95.10 97.62 96.34 e-adjp 384 373 332 89.01 86.46 87.71 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 810 717 88.52 87.23 87.87 i-advp 100 84 71 84.52 71.00 77.17 e-sbar 503 490 461 94.08 91.65 92.85 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.04 71.81 74.79 Avg2. 46451 46451 45291 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12249 11779 96.16 96.39 96.28 pp 4633 4664 4564 97.86 98.51 98.18 vp 4768 4735 4552 96.14 95.47 95.80 sbar 503 490 456 93.06 90.66 91.84 adjp 384 373 318 85.25 82.81 84.02 advp 822 810 711 87.78 86.50 87.13 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.80 77.09 79.84 Avg2. 23486 23471 22517 95.94 95.87 95.90 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 69 Log-likelihood = -86809.156873 Norm (log-likelihood gradient vector) = 3361.973775 Norm (lambda vector) = 197.989474 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4577 98.11 98.79 98.45 i-np 13660 13640 13371 98.03 97.88 97.96 e-np 12220 12254 12044 98.29 98.56 98.42 o 6349 6369 6191 97.21 97.51 97.36 e-vp 4768 4730 4658 98.48 97.69 98.08 i-vp 2602 2677 2541 94.92 97.66 96.27 e-adjp 384 375 333 88.80 86.72 87.75 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 810 719 88.77 87.47 88.11 i-advp 100 83 71 85.54 71.00 77.60 e-sbar 503 489 462 94.48 91.85 93.15 i-adjp 152 138 117 84.78 76.97 80.69 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.98 72.63 75.21 Avg2. 46451 46451 45283 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12254 11776 96.10 96.37 96.23 pp 4633 4665 4567 97.90 98.58 98.24 vp 4768 4730 4549 96.17 95.41 95.79 sbar 503 489 457 93.46 90.85 92.14 adjp 384 375 318 84.80 82.81 83.79 advp 822 810 713 88.02 86.74 87.38 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.82 77.76 80.21 Avg2. 23486 23474 22518 95.93 95.88 95.90 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 70 Log-likelihood = -84455.866722 Norm (log-likelihood gradient vector) = 3158.282769 Norm (lambda vector) = 200.788395 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4688 4582 97.74 98.90 98.32 i-np 13660 13814 13399 97.00 98.09 97.54 e-np 12220 12181 11982 98.37 98.05 98.21 o 6349 6292 6129 97.41 96.53 96.97 e-vp 4768 4726 4658 98.56 97.69 98.13 i-vp 2602 2664 2538 95.27 97.54 96.39 e-adjp 384 374 333 89.04 86.72 87.86 i-pp 52 40 34 85.00 65.38 73.91 e-advp 822 814 721 88.57 87.71 88.14 i-advp 100 83 71 85.54 71.00 77.60 e-sbar 503 463 444 95.90 88.27 91.93 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 13 10 76.92 83.33 80.00 i-conjp 24 15 15 100.00 62.50 76.92 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.21 71.74 74.84 Avg2. 46451 46451 45170 97.24 97.24 97.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12181 11671 95.81 95.51 95.66 pp 4633 4688 4572 97.53 98.68 98.10 vp 4768 4726 4552 96.32 95.47 95.89 sbar 503 463 440 95.03 87.48 91.10 adjp 384 374 318 85.03 82.81 83.91 advp 822 814 714 87.71 86.86 87.29 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.92 77.36 80.04 Avg2. 23486 23397 22405 95.76 95.40 95.58 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 71 Log-likelihood = -84857.093968 Norm (log-likelihood gradient vector) = 9918.506750 Norm (lambda vector) = 207.038381 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4677 4581 97.95 98.88 98.41 i-np 13660 13744 13404 97.53 98.13 97.83 e-np 12220 12210 12017 98.42 98.34 98.38 o 6349 6321 6162 97.48 97.05 97.27 e-vp 4768 4730 4659 98.50 97.71 98.10 i-vp 2602 2671 2539 95.06 97.58 96.30 e-adjp 384 372 334 89.78 86.98 88.36 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 814 722 88.70 87.83 88.26 i-advp 100 82 71 86.59 71.00 78.02 e-sbar 503 474 454 95.78 90.26 92.94 i-adjp 152 136 118 86.76 77.63 81.94 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.14 72.18 75.04 Avg2. 46451 46451 45259 97.43 97.43 97.43 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12210 11741 96.16 96.08 96.12 pp 4633 4677 4571 97.73 98.66 98.20 vp 4768 4730 4551 96.22 95.45 95.83 sbar 503 474 449 94.73 89.26 91.91 adjp 384 372 320 86.02 83.33 84.66 advp 822 814 715 87.84 86.98 87.41 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.04 77.66 80.26 Avg2. 23486 23428 22485 95.97 95.74 95.86 Current max chunk-based F1: 95.94 (iteration 62) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 72 Log-likelihood = -82963.201440 Norm (log-likelihood gradient vector) = 4692.239239 Norm (lambda vector) = 203.807906 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4669 4578 98.05 98.81 98.43 i-np 13660 13692 13411 97.95 98.18 98.06 e-np 12220 12236 12042 98.41 98.54 98.48 o 6349 6349 6192 97.53 97.53 97.53 e-vp 4768 4730 4662 98.56 97.78 98.17 i-vp 2602 2666 2542 95.35 97.69 96.51 e-adjp 384 374 334 89.30 86.98 88.13 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 815 723 88.71 87.96 88.33 i-advp 100 82 71 86.59 71.00 78.02 e-sbar 503 482 459 95.23 91.25 93.20 i-adjp 152 136 118 86.76 77.63 81.94 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 16 16 100.00 66.67 80.00 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.14 72.28 75.09 Avg2. 46451 46451 45330 97.59 97.59 97.59 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12236 11779 96.27 96.39 96.33 pp 4633 4669 4568 97.84 98.60 98.22 vp 4768 4730 4558 96.36 95.60 95.98 sbar 503 482 454 94.19 90.26 92.18 adjp 384 374 320 85.56 83.33 84.43 advp 822 815 717 87.98 87.23 87.60 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.99 77.82 80.32 Avg2. 23486 23457 22534 96.07 95.95 96.01 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 73 Log-likelihood = -81345.531821 Norm (log-likelihood gradient vector) = 2059.612598 Norm (lambda vector) = 205.254045 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4574 98.11 98.73 98.42 i-np 13660 13704 13410 97.85 98.17 98.01 e-np 12220 12230 12033 98.39 98.47 98.43 o 6349 6341 6183 97.51 97.39 97.45 e-vp 4768 4731 4661 98.52 97.76 98.14 i-vp 2602 2665 2541 95.35 97.66 96.49 e-adjp 384 379 336 88.65 87.50 88.07 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 810 721 89.01 87.71 88.36 i-advp 100 82 71 86.59 71.00 78.02 e-sbar 503 489 462 94.48 91.85 93.15 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.10 73.49 75.72 Avg2. 46451 46451 45311 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11763 96.18 96.26 96.22 pp 4633 4662 4565 97.92 98.53 98.22 vp 4768 4731 4558 96.34 95.60 95.97 sbar 503 489 457 93.46 90.85 92.14 adjp 384 379 322 84.96 83.85 84.40 advp 822 810 714 88.15 86.86 87.50 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.87 79.13 80.96 Avg2. 23486 23454 22519 96.01 95.88 95.95 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 74 Log-likelihood = -80594.316555 Norm (log-likelihood gradient vector) = 1940.026694 Norm (lambda vector) = 205.891519 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4576 98.05 98.77 98.41 i-np 13660 13737 13411 97.63 98.18 97.90 e-np 12220 12213 12017 98.40 98.34 98.37 o 6349 6322 6166 97.53 97.12 97.32 e-vp 4768 4732 4661 98.50 97.76 98.13 i-vp 2602 2665 2541 95.35 97.66 96.49 e-adjp 384 379 336 88.65 87.50 88.07 i-pp 52 38 34 89.47 65.38 75.56 e-advp 822 807 719 89.10 87.47 88.28 i-advp 100 82 71 86.59 71.00 78.02 e-sbar 503 488 461 94.47 91.65 93.04 i-adjp 152 135 116 85.93 76.32 80.84 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.14 73.45 75.72 Avg2. 46451 46451 45278 97.47 97.47 97.47 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12213 11735 96.09 96.03 96.06 pp 4633 4667 4568 97.88 98.60 98.24 vp 4768 4732 4557 96.30 95.57 95.94 sbar 503 488 456 93.44 90.66 92.03 adjp 384 379 321 84.70 83.59 84.14 advp 822 807 712 88.23 86.62 87.42 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.84 79.04 80.89 Avg2. 23486 23439 22489 95.95 95.75 95.85 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 75 Log-likelihood = -79531.798693 Norm (log-likelihood gradient vector) = 2828.511582 Norm (lambda vector) = 207.274976 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4570 98.09 98.64 98.36 i-np 13660 13561 13329 98.29 97.58 97.93 e-np 12220 12299 12062 98.07 98.71 98.39 o 6349 6416 6214 96.85 97.87 97.36 e-vp 4768 4736 4662 98.44 97.78 98.11 i-vp 2602 2661 2536 95.30 97.46 96.37 e-adjp 384 383 338 88.25 88.02 88.14 i-pp 52 36 34 94.44 65.38 77.27 e-advp 822 800 717 89.62 87.23 88.41 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 496 463 93.35 92.05 92.69 i-adjp 152 135 115 85.19 75.66 80.14 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 20 20 100.00 83.33 90.91 e-conjp 16 11 11 100.00 68.75 81.48 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.27 73.93 76.04 Avg2. 46451 46451 45281 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12299 11795 95.90 96.52 96.21 pp 4633 4659 4563 97.94 98.49 98.21 vp 4768 4736 4560 96.28 95.64 95.96 sbar 503 496 458 92.34 91.05 91.69 adjp 384 383 322 84.07 83.85 83.96 advp 822 800 711 88.88 86.50 87.67 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 11 11 100.00 68.75 81.48 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.71 79.76 81.21 Avg2. 23486 23527 22550 95.85 96.01 95.93 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 76 Log-likelihood = -78986.790838 Norm (log-likelihood gradient vector) = 6904.440120 Norm (lambda vector) = 210.211777 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4573 98.11 98.70 98.41 i-np 13660 13704 13407 97.83 98.15 97.99 e-np 12220 12230 12034 98.40 98.48 98.44 o 6349 6348 6187 97.46 97.45 97.46 e-vp 4768 4735 4664 98.50 97.82 98.16 i-vp 2602 2666 2540 95.27 97.62 96.43 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 36 34 94.44 65.38 77.27 e-advp 822 800 716 89.50 87.10 88.29 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 492 463 94.11 92.05 93.07 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.43 73.40 75.83 Avg2. 46451 46451 45309 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11770 96.24 96.32 96.28 pp 4633 4661 4566 97.96 98.55 98.26 vp 4768 4735 4561 96.33 95.66 95.99 sbar 503 492 458 93.09 91.05 92.06 adjp 384 379 322 84.96 83.85 84.40 advp 822 800 710 88.75 86.37 87.55 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.91 79.11 80.97 Avg2. 23486 23450 22527 96.06 95.92 95.99 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 77 Log-likelihood = -77584.364587 Norm (log-likelihood gradient vector) = 2455.523069 Norm (lambda vector) = 210.330236 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4575 98.09 98.75 98.42 i-np 13660 13739 13411 97.61 98.18 97.89 e-np 12220 12215 12017 98.38 98.34 98.36 o 6349 6329 6169 97.47 97.16 97.32 e-vp 4768 4734 4663 98.50 97.80 98.15 i-vp 2602 2667 2541 95.28 97.66 96.45 e-adjp 384 377 337 89.39 87.76 88.57 i-pp 52 36 34 94.44 65.38 77.27 e-advp 822 799 716 89.61 87.10 88.34 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 489 462 94.48 91.85 93.15 i-adjp 152 135 117 86.67 76.97 81.53 e-prt 126 133 122 91.73 96.83 94.21 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.48 73.44 75.87 Avg2. 46451 46451 45281 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12215 11743 96.14 96.10 96.12 pp 4633 4664 4568 97.94 98.60 98.27 vp 4768 4734 4561 96.35 95.66 96.00 sbar 503 489 457 93.46 90.85 92.14 adjp 384 377 323 85.68 84.11 84.89 advp 822 799 710 88.86 86.37 87.60 prt 126 133 122 91.73 96.83 94.21 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.01 79.10 81.01 Avg2. 23486 23431 22502 96.04 95.81 95.92 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 78 Log-likelihood = -76944.830986 Norm (log-likelihood gradient vector) = 2154.841578 Norm (lambda vector) = 210.666081 Log-likelihood and gradient computational time: 323 seconds Training iteration elapsed: 323 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4576 98.11 98.77 98.44 i-np 13660 13755 13419 97.56 98.24 97.90 e-np 12220 12206 12013 98.42 98.31 98.36 o 6349 6319 6163 97.53 97.07 97.30 e-vp 4768 4735 4664 98.50 97.82 98.16 i-vp 2602 2668 2540 95.20 97.62 96.39 e-adjp 384 372 336 90.32 87.50 88.89 i-pp 52 37 34 91.89 65.38 76.40 e-advp 822 804 717 89.18 87.23 88.19 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 487 462 94.87 91.85 93.33 i-adjp 152 138 117 84.78 76.97 80.69 e-prt 126 132 122 92.42 96.83 94.57 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.33 73.43 75.80 Avg2. 46451 46451 45280 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12206 11742 96.20 96.09 96.14 pp 4633 4664 4568 97.94 98.60 98.27 vp 4768 4735 4558 96.26 95.60 95.93 sbar 503 487 457 93.84 90.85 92.32 adjp 384 372 321 86.29 83.59 84.92 advp 822 804 711 88.43 86.50 87.45 prt 126 132 122 92.42 96.83 94.57 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.14 79.06 81.05 Avg2. 23486 23420 22497 96.06 95.79 95.92 Current max chunk-based F1: 96.01 (iteration 72) Training iteration elapsed (including evaluation time): 357 seconds Iteration: 79 Log-likelihood = -75486.954655 Norm (log-likelihood gradient vector) = 2301.732091 Norm (lambda vector) = 212.623838 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4683 4583 97.86 98.92 98.39 i-np 13660 13672 13401 98.02 98.10 98.06 e-np 12220 12249 12046 98.34 98.58 98.46 o 6349 6356 6195 97.47 97.57 97.52 e-vp 4768 4730 4662 98.56 97.78 98.17 i-vp 2602 2670 2542 95.21 97.69 96.43 e-adjp 384 371 335 90.30 87.24 88.74 i-pp 52 37 35 94.59 67.31 78.65 e-advp 822 812 723 89.04 87.96 88.49 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 473 453 95.77 90.06 92.83 i-adjp 152 140 119 85.00 78.29 81.51 e-prt 126 131 122 93.13 96.83 94.94 i-sbar 12 13 10 76.92 83.33 80.00 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.73 71.59 74.99 Avg2. 46451 46451 45325 97.58 97.58 97.58 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12249 11791 96.26 96.49 96.38 pp 4633 4683 4577 97.74 98.79 98.26 vp 4768 4730 4558 96.36 95.60 95.98 sbar 503 473 449 94.93 89.26 92.01 adjp 384 371 321 86.52 83.59 85.03 advp 822 812 717 88.30 87.23 87.76 prt 126 131 122 93.13 96.83 94.94 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.32 77.15 80.12 Avg2. 23486 23466 22550 96.10 96.01 96.06 Current max chunk-based F1: 96.06 (iteration 79) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 80 Log-likelihood = -74204.445653 Norm (log-likelihood gradient vector) = 3837.933311 Norm (lambda vector) = 214.767864 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4680 4581 97.88 98.88 98.38 i-np 13660 13703 13422 97.95 98.26 98.10 e-np 12220 12228 12038 98.45 98.51 98.48 o 6349 6340 6187 97.59 97.45 97.52 e-vp 4768 4734 4666 98.56 97.86 98.21 i-vp 2602 2664 2539 95.31 97.58 96.43 e-adjp 384 372 335 90.05 87.24 88.62 i-pp 52 39 35 89.74 67.31 76.92 e-advp 822 811 721 88.90 87.71 88.30 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 479 456 95.20 90.66 92.87 i-adjp 152 144 123 85.42 80.92 83.11 e-prt 126 130 121 93.08 96.03 94.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 13 13 100.00 54.17 70.27 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.19 71.48 74.69 Avg2. 46451 46451 45332 97.59 97.59 97.59 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11788 96.40 96.46 96.43 pp 4633 4680 4575 97.76 98.75 98.25 vp 4768 4734 4562 96.37 95.68 96.02 sbar 503 479 451 94.15 89.66 91.85 adjp 384 372 322 86.56 83.85 85.19 advp 822 811 715 88.16 86.98 87.57 prt 126 130 121 93.08 96.03 94.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.25 77.12 80.07 Avg2. 23486 23451 22549 96.15 96.01 96.08 Current max chunk-based F1: 96.08 (iteration 80) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 81 Log-likelihood = -72675.182141 Norm (log-likelihood gradient vector) = 2261.277234 Norm (lambda vector) = 217.419412 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4669 4575 97.99 98.75 98.37 i-np 13660 13686 13414 98.01 98.20 98.11 e-np 12220 12233 12041 98.43 98.54 98.48 o 6349 6353 6192 97.47 97.53 97.50 e-vp 4768 4735 4665 98.52 97.84 98.18 i-vp 2602 2662 2536 95.27 97.46 96.35 e-adjp 384 373 335 89.81 87.24 88.51 i-pp 52 39 35 89.74 67.31 76.92 e-advp 822 810 720 88.89 87.59 88.24 i-advp 100 82 70 85.37 70.00 76.92 e-sbar 503 489 459 93.87 91.25 92.54 i-adjp 152 144 122 84.72 80.26 82.43 e-prt 126 130 121 93.08 96.03 94.53 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 13 13 100.00 54.17 70.27 e-conjp 16 7 7 100.00 43.75 60.87 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.00 71.88 74.81 Avg2. 46451 46451 45324 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12233 11790 96.38 96.48 96.43 pp 4633 4669 4569 97.86 98.62 98.24 vp 4768 4735 4558 96.26 95.60 95.93 sbar 503 489 454 92.84 90.26 91.53 adjp 384 373 321 86.06 83.59 84.81 advp 822 810 714 88.15 86.86 87.50 prt 126 130 121 93.08 96.03 94.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 7 7 100.00 43.75 60.87 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.06 77.12 79.98 Avg2. 23486 23456 22542 96.10 95.98 96.04 Current max chunk-based F1: 96.08 (iteration 80) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 82 Log-likelihood = -71741.273557 Norm (log-likelihood gradient vector) = 1751.783095 Norm (lambda vector) = 219.010623 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4574 98.07 98.73 98.40 i-np 13660 13672 13407 98.06 98.15 98.10 e-np 12220 12239 12043 98.40 98.55 98.47 o 6349 6366 6200 97.39 97.65 97.52 e-vp 4768 4736 4667 98.54 97.88 98.21 i-vp 2602 2655 2537 95.56 97.50 96.52 e-adjp 384 374 336 89.84 87.50 88.65 i-pp 52 37 35 94.59 67.31 78.65 e-advp 822 814 724 88.94 88.08 88.51 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 492 462 93.90 91.85 92.86 i-adjp 152 143 123 86.01 80.92 83.39 e-prt 126 129 121 93.80 96.03 94.90 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 15 15 100.00 62.50 76.92 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.41 72.72 75.46 Avg2. 46451 46451 45341 97.61 97.61 97.61 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12239 11794 96.36 96.51 96.44 pp 4633 4664 4568 97.94 98.60 98.27 vp 4768 4736 4564 96.37 95.72 96.04 sbar 503 492 457 92.89 90.85 91.86 adjp 384 374 323 86.36 84.11 85.22 advp 822 814 718 88.21 87.35 87.78 prt 126 129 121 93.80 96.03 94.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.19 77.92 80.47 Avg2. 23486 23466 22561 96.14 96.06 96.10 Current max chunk-based F1: 96.10 (iteration 82) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 83 Log-likelihood = -70996.236714 Norm (log-likelihood gradient vector) = 1956.038068 Norm (lambda vector) = 220.606435 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4651 4567 98.19 98.58 98.38 i-np 13660 13698 13411 97.90 98.18 98.04 e-np 12220 12226 12035 98.44 98.49 98.46 o 6349 6344 6183 97.46 97.39 97.42 e-vp 4768 4736 4665 98.50 97.84 98.17 i-vp 2602 2658 2534 95.33 97.39 96.35 e-adjp 384 376 336 89.36 87.50 88.42 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 810 723 89.26 87.96 88.60 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 508 469 92.32 93.24 92.78 i-adjp 152 143 121 84.62 79.61 82.03 e-prt 126 130 122 93.85 96.83 95.31 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 17 17 100.00 70.83 82.93 e-conjp 16 9 9 100.00 56.25 72.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.13 73.45 75.72 Avg2. 46451 46451 45316 97.56 97.56 97.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12226 11776 96.32 96.37 96.34 pp 4633 4651 4560 98.04 98.42 98.23 vp 4768 4736 4556 96.20 95.55 95.88 sbar 503 508 464 91.34 92.25 91.79 adjp 384 376 320 85.11 83.33 84.21 advp 822 810 717 88.52 87.23 87.87 prt 126 130 122 93.85 96.83 95.31 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 9 9 100.00 56.25 72.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.94 78.62 80.72 Avg2. 23486 23456 22532 96.06 95.94 96.00 Current max chunk-based F1: 96.10 (iteration 82) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 84 Log-likelihood = -69784.926449 Norm (log-likelihood gradient vector) = 3248.417910 Norm (lambda vector) = 223.492004 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4571 98.11 98.66 98.39 i-np 13660 13613 13374 98.24 97.91 98.08 e-np 12220 12264 12057 98.31 98.67 98.49 o 6349 6386 6204 97.15 97.72 97.43 e-vp 4768 4737 4664 98.46 97.82 98.14 i-vp 2602 2657 2532 95.30 97.31 96.29 e-adjp 384 380 338 88.95 88.02 88.48 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 809 724 89.49 88.08 88.78 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 500 467 93.40 92.84 93.12 i-adjp 152 142 122 85.92 80.26 82.99 e-prt 126 130 122 93.85 96.83 95.31 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.22 74.24 76.18 Avg2. 46451 46451 45328 97.58 97.58 97.58 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12264 11801 96.22 96.57 96.40 pp 4633 4659 4564 97.96 98.51 98.24 vp 4768 4737 4554 96.14 95.51 95.82 sbar 503 500 462 92.40 91.85 92.12 adjp 384 380 324 85.26 84.38 84.82 advp 822 809 718 88.75 87.35 88.04 prt 126 130 122 93.85 96.83 95.31 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.06 79.35 81.16 Avg2. 23486 23499 22563 96.02 96.07 96.04 Current max chunk-based F1: 96.10 (iteration 82) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 85 Log-likelihood = -68752.893424 Norm (log-likelihood gradient vector) = 4350.745748 Norm (lambda vector) = 226.546714 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4576 98.07 98.77 98.42 i-np 13660 13749 13434 97.71 98.35 98.03 e-np 12220 12199 12024 98.57 98.40 98.48 o 6349 6325 6174 97.61 97.24 97.43 e-vp 4768 4737 4662 98.42 97.78 98.10 i-vp 2602 2655 2527 95.18 97.12 96.14 e-adjp 384 376 336 89.36 87.50 88.42 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 807 723 89.59 87.96 88.77 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 493 463 93.91 92.05 92.97 i-adjp 152 140 121 86.43 79.61 82.88 e-prt 126 130 122 93.85 96.83 95.31 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.30 74.11 76.15 Avg2. 46451 46451 45315 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12199 11766 96.45 96.28 96.37 pp 4633 4666 4569 97.92 98.62 98.27 vp 4768 4737 4547 95.99 95.36 95.68 sbar 503 493 458 92.90 91.05 91.97 adjp 384 376 321 85.37 83.59 84.47 advp 822 807 717 88.85 87.23 88.03 prt 126 130 122 93.85 96.83 95.31 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.13 79.15 81.09 Avg2. 23486 23428 22518 96.12 95.88 96.00 Current max chunk-based F1: 96.10 (iteration 82) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 86 Log-likelihood = -68000.613495 Norm (log-likelihood gradient vector) = 3597.882983 Norm (lambda vector) = 228.732506 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4577 98.11 98.79 98.45 i-np 13660 13694 13412 97.94 98.18 98.06 e-np 12220 12226 12040 98.48 98.53 98.50 o 6349 6349 6189 97.48 97.48 97.48 e-vp 4768 4739 4665 98.44 97.84 98.14 i-vp 2602 2658 2531 95.22 97.27 96.24 e-adjp 384 376 336 89.36 87.50 88.42 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 809 724 89.49 88.08 88.78 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 492 463 94.11 92.05 93.07 i-adjp 152 139 121 87.05 79.61 83.16 e-prt 126 130 122 93.85 96.83 95.31 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.34 74.14 76.18 Avg2. 46451 46451 45333 97.59 97.59 97.59 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12226 11786 96.40 96.45 96.42 pp 4633 4665 4570 97.96 98.64 98.30 vp 4768 4739 4554 96.10 95.51 95.80 sbar 503 492 458 93.09 91.05 92.06 adjp 384 376 322 85.64 83.85 84.74 advp 822 809 718 88.75 87.35 88.04 prt 126 130 122 93.85 96.83 95.31 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.18 79.22 81.15 Avg2. 23486 23457 22548 96.12 96.01 96.07 Current max chunk-based F1: 96.10 (iteration 82) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 87 Log-likelihood = -67434.925365 Norm (log-likelihood gradient vector) = 1767.859843 Norm (lambda vector) = 228.080443 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4576 98.11 98.77 98.44 i-np 13660 13679 13414 98.06 98.20 98.13 e-np 12220 12240 12049 98.44 98.60 98.52 o 6349 6358 6195 97.44 97.57 97.51 e-vp 4768 4738 4666 98.48 97.86 98.17 i-vp 2602 2659 2534 95.30 97.39 96.33 e-adjp 384 374 336 89.84 87.50 88.65 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 809 726 89.74 88.32 89.03 i-advp 100 81 70 86.42 70.00 77.35 e-sbar 503 493 464 94.12 92.25 93.17 i-adjp 152 132 116 87.88 76.32 81.69 e-prt 126 131 122 93.13 96.83 94.94 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 19 19 100.00 79.17 88.37 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.39 74.01 76.14 Avg2. 46451 46451 45351 97.63 97.63 97.63 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12240 11801 96.41 96.57 96.49 pp 4633 4664 4569 97.96 98.62 98.29 vp 4768 4738 4557 96.18 95.57 95.88 sbar 503 493 459 93.10 91.25 92.17 adjp 384 374 322 86.10 83.85 84.96 advp 822 809 720 89.00 87.59 88.29 prt 126 131 122 93.13 96.83 94.94 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.19 79.28 81.19 Avg2. 23486 23469 22568 96.16 96.09 96.13 Current max chunk-based F1: 96.13 (iteration 87) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 88 Log-likelihood = -66896.963388 Norm (log-likelihood gradient vector) = 1712.111979 Norm (lambda vector) = 228.344637 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4576 98.09 98.77 98.43 i-np 13660 13669 13407 98.08 98.15 98.12 e-np 12220 12239 12049 98.45 98.60 98.52 o 6349 6362 6198 97.42 97.62 97.52 e-vp 4768 4738 4666 98.48 97.86 98.17 i-vp 2602 2655 2534 95.44 97.39 96.40 e-adjp 384 372 336 90.32 87.50 88.89 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 815 727 89.20 88.44 88.82 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 494 465 94.13 92.45 93.28 i-adjp 152 136 120 88.24 78.95 83.33 e-prt 126 131 122 93.13 96.83 94.94 i-sbar 12 16 11 68.75 91.67 78.57 i-conjp 24 18 18 100.00 75.00 85.71 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.31 73.95 76.07 Avg2. 46451 46451 45352 97.63 97.63 97.63 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12239 11798 96.40 96.55 96.47 pp 4633 4665 4569 97.94 98.62 98.28 vp 4768 4738 4560 96.24 95.64 95.94 sbar 503 494 460 93.12 91.45 92.28 adjp 384 372 323 86.83 84.11 85.45 advp 822 815 721 88.47 87.71 88.09 prt 126 131 122 93.13 96.83 94.94 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.21 79.34 81.23 Avg2. 23486 23474 22571 96.15 96.10 96.13 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 89 Log-likelihood = -65860.148147 Norm (log-likelihood gradient vector) = 2084.415795 Norm (lambda vector) = 229.831466 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4574 98.22 98.73 98.47 i-np 13660 13723 13411 97.73 98.18 97.95 e-np 12220 12209 12026 98.50 98.41 98.46 o 6349 6338 6177 97.46 97.29 97.38 e-vp 4768 4736 4663 98.46 97.80 98.13 i-vp 2602 2655 2533 95.40 97.35 96.37 e-adjp 384 377 335 88.86 87.24 88.04 i-pp 52 38 35 92.11 67.31 77.78 e-advp 822 808 720 89.11 87.59 88.34 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 500 469 93.80 93.24 93.52 i-adjp 152 142 122 85.92 80.26 82.99 e-prt 126 131 122 93.13 96.83 94.94 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 18 18 100.00 75.00 85.71 e-conjp 16 10 10 100.00 62.50 76.92 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.27 73.97 76.06 Avg2. 46451 46451 45304 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12209 11760 96.32 96.24 96.28 pp 4633 4657 4567 98.07 98.58 98.32 vp 4768 4736 4558 96.24 95.60 95.92 sbar 503 500 464 92.80 92.25 92.52 adjp 384 377 323 85.68 84.11 84.89 advp 822 808 714 88.37 86.86 87.61 prt 126 131 122 93.13 96.83 94.94 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 10 10 100.00 62.50 76.92 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.06 79.30 81.13 Avg2. 23486 23438 22526 96.11 95.91 96.01 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 90 Log-likelihood = -64015.685158 Norm (log-likelihood gradient vector) = 2648.994121 Norm (lambda vector) = 233.632882 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4682 4581 97.84 98.88 98.36 i-np 13660 13651 13377 97.99 97.93 97.96 e-np 12220 12243 12042 98.36 98.54 98.45 o 6349 6360 6188 97.30 97.46 97.38 e-vp 4768 4731 4659 98.48 97.71 98.09 i-vp 2602 2659 2538 95.45 97.54 96.48 e-adjp 384 377 334 88.59 86.98 87.78 i-pp 52 37 35 94.59 67.31 78.65 e-advp 822 812 721 88.79 87.71 88.25 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 485 459 94.64 91.25 92.91 i-adjp 152 149 125 83.89 82.24 83.06 e-prt 126 132 123 93.18 97.62 95.35 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.25 72.61 75.32 Avg2. 46451 46451 45294 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12243 11776 96.19 96.37 96.28 pp 4633 4682 4574 97.69 98.73 98.21 vp 4768 4731 4558 96.34 95.60 95.97 sbar 503 485 454 93.61 90.26 91.90 adjp 384 377 321 85.15 83.59 84.36 advp 822 812 715 88.05 86.98 87.52 prt 126 132 123 93.18 97.62 95.35 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.02 77.91 80.39 Avg2. 23486 23480 22537 95.98 95.96 95.97 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 91 Log-likelihood = -62524.261972 Norm (log-likelihood gradient vector) = 4262.793102 Norm (lambda vector) = 238.964987 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4578 98.09 98.81 98.45 i-np 13660 13675 13398 97.97 98.08 98.03 e-np 12220 12230 12038 98.43 98.51 98.47 o 6349 6356 6190 97.39 97.50 97.44 e-vp 4768 4732 4658 98.44 97.69 98.06 i-vp 2602 2652 2530 95.40 97.23 96.31 e-adjp 384 377 333 88.33 86.72 87.52 i-pp 52 37 35 94.59 67.31 78.65 e-advp 822 812 721 88.79 87.71 88.25 i-advp 100 88 71 80.68 71.00 75.53 e-sbar 503 496 467 94.15 92.84 93.49 i-adjp 152 150 126 84.00 82.89 83.44 e-prt 126 132 123 93.18 97.62 95.35 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 14 14 100.00 58.33 73.68 e-conjp 16 8 8 100.00 50.00 66.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.14 72.70 75.32 Avg2. 46451 46451 45309 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11776 96.29 96.37 96.33 pp 4633 4667 4571 97.94 98.66 98.30 vp 4768 4732 4553 96.22 95.49 95.85 sbar 503 496 462 93.15 91.85 92.49 adjp 384 377 321 85.15 83.59 84.36 advp 822 812 715 88.05 86.98 87.52 prt 126 132 123 93.18 97.62 95.35 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 8 8 100.00 50.00 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 83.00 78.06 80.45 Avg2. 23486 23464 22537 96.05 95.96 96.00 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 92 Log-likelihood = -61653.415077 Norm (log-likelihood gradient vector) = 1652.129138 Norm (lambda vector) = 238.653884 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4579 98.20 98.83 98.52 i-np 13660 13718 13412 97.77 98.18 97.98 e-np 12220 12210 12029 98.52 98.44 98.48 o 6349 6333 6183 97.63 97.39 97.51 e-vp 4768 4733 4660 98.46 97.73 98.09 i-vp 2602 2650 2530 95.47 97.23 96.34 e-adjp 384 377 333 88.33 86.72 87.52 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 806 718 89.08 87.35 88.21 i-advp 100 83 69 83.13 69.00 75.41 e-sbar 503 499 469 93.99 93.24 93.61 i-adjp 152 150 126 84.00 82.89 83.44 e-prt 126 133 123 92.48 97.62 94.98 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.60 74.88 76.22 Avg2. 46451 46451 45317 97.56 97.56 97.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12210 11768 96.38 96.30 96.34 pp 4633 4663 4572 98.05 98.68 98.36 vp 4768 4733 4556 96.26 95.55 95.91 sbar 503 499 464 92.99 92.25 92.61 adjp 384 377 321 85.15 83.59 84.36 advp 822 806 712 88.34 86.62 87.47 prt 126 133 123 92.48 97.62 94.98 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.13 79.94 81.02 Avg2. 23486 23443 22535 96.13 95.95 96.04 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 93 Log-likelihood = -61163.805252 Norm (log-likelihood gradient vector) = 1539.842173 Norm (lambda vector) = 239.007183 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4578 98.20 98.81 98.50 i-np 13660 13728 13415 97.72 98.21 97.96 e-np 12220 12208 12024 98.49 98.40 98.44 o 6349 6329 6180 97.65 97.34 97.49 e-vp 4768 4732 4659 98.46 97.71 98.08 i-vp 2602 2652 2529 95.36 97.19 96.27 e-adjp 384 375 333 88.80 86.72 87.75 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 804 716 89.05 87.10 88.07 i-advp 100 83 69 83.13 69.00 75.41 e-sbar 503 501 469 93.61 93.24 93.43 i-adjp 152 148 126 85.14 82.89 84.00 e-prt 126 133 123 92.48 97.62 94.98 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.65 74.86 76.23 Avg2. 46451 46451 45307 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12208 11760 96.33 96.24 96.28 pp 4633 4662 4571 98.05 98.66 98.35 vp 4768 4732 4550 96.15 95.43 95.79 sbar 503 501 464 92.61 92.25 92.43 adjp 384 375 321 85.60 83.59 84.58 advp 822 804 710 88.31 86.37 87.33 prt 126 133 123 92.48 97.62 94.98 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.12 79.89 80.99 Avg2. 23486 23437 22518 96.08 95.88 95.98 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 94 Log-likelihood = -60594.430430 Norm (log-likelihood gradient vector) = 2151.084348 Norm (lambda vector) = 240.071131 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4654 4574 98.28 98.73 98.50 i-np 13660 13651 13381 98.02 97.96 97.99 e-np 12220 12244 12046 98.38 98.58 98.48 o 6349 6373 6202 97.32 97.68 97.50 e-vp 4768 4739 4663 98.40 97.80 98.10 i-vp 2602 2640 2523 95.57 96.96 96.26 e-adjp 384 377 334 88.59 86.98 87.78 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 806 715 88.71 86.98 87.84 i-advp 100 83 69 83.13 69.00 75.41 e-sbar 503 504 469 93.06 93.24 93.15 i-adjp 152 152 126 82.89 82.89 82.89 e-prt 126 133 123 92.48 97.62 94.98 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.40 74.45 75.90 Avg2. 46451 46451 45310 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12244 11787 96.27 96.46 96.36 pp 4633 4654 4567 98.13 98.58 98.35 vp 4768 4739 4556 96.14 95.55 95.85 sbar 503 504 463 91.87 92.05 91.96 adjp 384 377 322 85.41 83.85 84.63 advp 822 806 709 87.97 86.25 87.10 prt 126 133 123 92.48 97.62 94.98 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.99 79.91 80.94 Avg2. 23486 23479 22546 96.03 96.00 96.01 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 95 Log-likelihood = -59897.820268 Norm (log-likelihood gradient vector) = 2910.317669 Norm (lambda vector) = 242.571550 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4654 4575 98.30 98.75 98.52 i-np 13660 13759 13415 97.50 98.21 97.85 e-np 12220 12197 12014 98.50 98.31 98.41 o 6349 6328 6170 97.50 97.18 97.34 e-vp 4768 4736 4664 98.48 97.82 98.15 i-vp 2602 2642 2526 95.61 97.08 96.34 e-adjp 384 372 334 89.78 86.98 88.36 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 808 717 88.74 87.23 87.98 i-advp 100 85 69 81.18 69.00 74.59 e-sbar 503 498 467 93.78 92.84 93.31 i-adjp 152 145 122 84.14 80.26 82.15 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 13 10 76.92 83.33 80.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.77 74.34 76.01 Avg2. 46451 46451 45282 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12197 11738 96.24 96.06 96.15 pp 4633 4654 4568 98.15 98.60 98.37 vp 4768 4736 4558 96.24 95.60 95.92 sbar 503 498 462 92.77 91.85 92.31 adjp 384 372 323 86.83 84.11 85.45 advp 822 808 711 88.00 86.50 87.24 prt 126 133 124 93.23 98.41 95.75 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.31 79.99 81.13 Avg2. 23486 23420 22503 96.08 95.81 95.95 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 96 Log-likelihood = -58763.588201 Norm (log-likelihood gradient vector) = 2646.852706 Norm (lambda vector) = 246.005970 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4656 4576 98.28 98.77 98.53 i-np 13660 13725 13412 97.72 98.18 97.95 e-np 12220 12214 12025 98.45 98.40 98.43 o 6349 6344 6182 97.45 97.37 97.41 e-vp 4768 4736 4665 98.50 97.84 98.17 i-vp 2602 2645 2530 95.65 97.23 96.44 e-adjp 384 370 334 90.27 86.98 88.59 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 811 720 88.78 87.59 88.18 i-advp 100 85 69 81.18 69.00 74.59 e-sbar 503 496 466 93.95 92.64 93.29 i-adjp 152 143 122 85.31 80.26 82.71 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 10 83.33 83.33 83.33 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.19 74.37 76.23 Avg2. 46451 46451 45310 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12214 11758 96.27 96.22 96.24 pp 4633 4656 4569 98.13 98.62 98.37 vp 4768 4736 4560 96.28 95.64 95.96 sbar 503 496 462 93.15 91.85 92.49 adjp 384 370 323 87.30 84.11 85.68 advp 822 811 714 88.04 86.86 87.45 prt 126 133 124 93.23 98.41 95.75 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.41 80.05 81.21 Avg2. 23486 23438 22529 96.12 95.93 96.02 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 97 Log-likelihood = -58084.835827 Norm (log-likelihood gradient vector) = 1683.008260 Norm (lambda vector) = 246.830383 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4577 98.22 98.79 98.50 i-np 13660 13703 13399 97.78 98.09 97.94 e-np 12220 12223 12029 98.41 98.44 98.42 o 6349 6354 6186 97.36 97.43 97.39 e-vp 4768 4734 4663 98.50 97.80 98.15 i-vp 2602 2648 2531 95.58 97.27 96.42 e-adjp 384 373 334 89.54 86.98 88.24 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 813 720 88.56 87.59 88.07 i-advp 100 84 69 82.14 69.00 75.00 e-sbar 503 494 465 94.13 92.45 93.28 i-adjp 152 139 122 87.77 80.26 83.85 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 10 83.33 83.33 83.33 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.31 74.36 76.28 Avg2. 46451 46451 45304 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12223 11760 96.21 96.24 96.22 pp 4633 4660 4570 98.07 98.64 98.35 vp 4768 4734 4559 96.30 95.62 95.96 sbar 503 494 461 93.32 91.65 92.48 adjp 384 373 324 86.86 84.38 85.60 advp 822 813 714 87.82 86.86 87.34 prt 126 133 124 93.23 98.41 95.75 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.35 80.05 81.19 Avg2. 23486 23452 22531 96.07 95.93 96.00 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 98 Log-likelihood = -57109.076380 Norm (log-likelihood gradient vector) = 1672.441615 Norm (lambda vector) = 248.660813 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4578 98.20 98.81 98.50 i-np 13660 13696 13398 97.82 98.08 97.95 e-np 12220 12226 12032 98.41 98.46 98.44 o 6349 6353 6186 97.37 97.43 97.40 e-vp 4768 4737 4662 98.42 97.78 98.10 i-vp 2602 2656 2536 95.48 97.46 96.46 e-adjp 384 372 334 89.78 86.98 88.36 i-pp 52 38 36 94.74 69.23 80.00 e-advp 822 810 720 88.89 87.59 88.24 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 492 464 94.31 92.25 93.27 i-adjp 152 138 121 87.68 79.61 83.45 e-prt 126 133 124 93.23 98.41 95.75 i-sbar 12 12 10 83.33 83.33 83.33 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.45 74.38 76.36 Avg2. 46451 46451 45310 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12226 11764 96.22 96.27 96.24 pp 4633 4662 4571 98.05 98.66 98.35 vp 4768 4737 4558 96.22 95.60 95.91 sbar 503 492 460 93.50 91.45 92.46 adjp 384 372 324 87.10 84.38 85.71 advp 822 810 714 88.15 86.86 87.50 prt 126 133 124 93.23 98.41 95.75 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.41 80.04 81.21 Avg2. 23486 23454 22534 96.08 95.95 96.01 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 99 Log-likelihood = -56335.410719 Norm (log-likelihood gradient vector) = 1687.010755 Norm (lambda vector) = 250.401559 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4654 4572 98.24 98.68 98.46 i-np 13660 13638 13366 98.01 97.85 97.93 e-np 12220 12251 12040 98.28 98.53 98.40 o 6349 6373 6193 97.18 97.54 97.36 e-vp 4768 4734 4662 98.48 97.78 98.13 i-vp 2602 2657 2538 95.52 97.54 96.52 e-adjp 384 378 338 89.42 88.02 88.71 i-pp 52 40 36 90.00 69.23 78.26 e-advp 822 809 717 88.63 87.23 87.92 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 502 468 93.23 93.04 93.13 i-adjp 152 141 122 86.52 80.26 83.28 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 74.63 76.01 Avg2. 46451 46451 45297 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12251 11775 96.11 96.36 96.24 pp 4633 4654 4564 98.07 98.51 98.29 vp 4768 4734 4560 96.32 95.64 95.98 sbar 503 502 461 91.83 91.65 91.74 adjp 384 378 326 86.24 84.90 85.56 advp 822 809 710 87.76 86.37 87.06 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.33 80.68 81.50 Avg2. 23486 23482 22540 95.99 95.97 95.98 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 100 Log-likelihood = -54989.629242 Norm (log-likelihood gradient vector) = 3511.364339 Norm (lambda vector) = 254.659016 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4577 98.05 98.79 98.42 i-np 13660 13720 13410 97.74 98.17 97.95 e-np 12220 12215 12028 98.47 98.43 98.45 o 6349 6328 6171 97.52 97.20 97.36 e-vp 4768 4734 4662 98.48 97.78 98.13 i-vp 2602 2647 2530 95.58 97.23 96.40 e-adjp 384 377 338 89.66 88.02 88.83 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 811 721 88.90 87.71 88.30 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 488 460 94.26 91.45 92.84 i-adjp 152 147 122 82.99 80.26 81.61 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.98 74.56 75.75 Avg2. 46451 46451 45300 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12215 11765 96.32 96.28 96.30 pp 4633 4668 4568 97.86 98.60 98.23 vp 4768 4734 4558 96.28 95.60 95.94 sbar 503 488 454 93.03 90.26 91.62 adjp 384 377 324 85.94 84.38 85.15 advp 822 811 714 88.04 86.86 87.45 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.78 80.54 81.16 Avg2. 23486 23448 22527 96.07 95.92 95.99 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 101 Log-likelihood = -53678.559616 Norm (log-likelihood gradient vector) = 1630.129799 Norm (lambda vector) = 257.643188 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4669 4578 98.05 98.81 98.43 i-np 13660 13717 13407 97.74 98.15 97.94 e-np 12220 12215 12027 98.46 98.42 98.44 o 6349 6333 6173 97.47 97.23 97.35 e-vp 4768 4735 4662 98.46 97.78 98.12 i-vp 2602 2643 2526 95.57 97.08 96.32 e-adjp 384 377 338 89.66 88.02 88.83 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 813 721 88.68 87.71 88.20 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 487 460 94.46 91.45 92.93 i-adjp 152 147 122 82.99 80.26 81.61 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.47 74.55 75.98 Avg2. 46451 46451 45295 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12215 11763 96.30 96.26 96.28 pp 4633 4669 4569 97.86 98.62 98.24 vp 4768 4735 4558 96.26 95.60 95.93 sbar 503 487 454 93.22 90.26 91.72 adjp 384 377 324 85.94 84.38 85.15 advp 822 813 714 87.82 86.86 87.34 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.44 80.54 81.48 Avg2. 23486 23450 22526 96.06 95.91 95.99 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 102 Log-likelihood = -53291.789142 Norm (log-likelihood gradient vector) = 1327.504011 Norm (lambda vector) = 257.555527 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4578 98.18 98.81 98.49 i-np 13660 13736 13416 97.67 98.21 97.94 e-np 12220 12210 12022 98.46 98.38 98.42 o 6349 6324 6169 97.55 97.16 97.36 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2642 2524 95.53 97.00 96.26 e-adjp 384 374 335 89.57 87.24 88.39 i-pp 52 42 36 85.71 69.23 76.60 e-advp 822 811 720 88.78 87.59 88.18 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 494 466 94.33 92.64 93.48 i-adjp 152 146 122 83.56 80.26 81.88 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.10 74.60 75.83 Avg2. 46451 46451 45295 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12210 11759 96.31 96.23 96.27 pp 4633 4663 4568 97.96 98.60 98.28 vp 4768 4733 4555 96.24 95.53 95.88 sbar 503 494 459 92.91 91.25 92.08 adjp 384 374 322 86.10 83.85 84.96 advp 822 811 713 87.92 86.74 87.32 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.44 80.64 81.53 Avg2. 23486 23440 22521 96.08 95.89 95.99 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 103 Log-likelihood = -52239.292885 Norm (log-likelihood gradient vector) = 1887.912857 Norm (lambda vector) = 259.041924 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4673 4582 98.05 98.90 98.47 i-np 13660 13609 13354 98.13 97.76 97.94 e-np 12220 12271 12051 98.21 98.62 98.41 o 6349 6386 6200 97.09 97.65 97.37 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2635 2520 95.64 96.85 96.24 e-adjp 384 375 335 89.33 87.24 88.27 i-pp 52 42 36 85.71 69.23 76.60 e-advp 822 820 723 88.17 87.96 88.06 i-advp 100 87 71 81.61 71.00 75.94 e-sbar 503 485 460 94.85 91.45 93.12 i-adjp 152 147 122 82.99 80.26 81.61 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 11 9 81.82 75.00 78.26 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.63 74.57 76.07 Avg2. 46451 46451 45290 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12271 11790 96.08 96.48 96.28 pp 4633 4673 4572 97.84 98.68 98.26 vp 4768 4733 4555 96.24 95.53 95.88 sbar 503 485 455 93.81 90.46 92.11 adjp 384 375 322 85.87 83.85 84.85 advp 822 820 716 87.32 87.10 87.21 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.42 80.63 81.51 Avg2. 23486 23512 22555 95.93 96.04 95.98 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 104 Log-likelihood = -51850.841179 Norm (log-likelihood gradient vector) = 4413.776710 Norm (lambda vector) = 262.158417 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4579 98.14 98.83 98.48 i-np 13660 13696 13399 97.83 98.09 97.96 e-np 12220 12234 12034 98.37 98.48 98.42 o 6349 6346 6180 97.38 97.34 97.36 e-vp 4768 4734 4664 98.52 97.82 98.17 i-vp 2602 2634 2521 95.71 96.89 96.29 e-adjp 384 373 335 89.81 87.24 88.51 i-pp 52 42 36 85.71 69.23 76.60 e-advp 822 816 722 88.48 87.83 88.16 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 490 464 94.69 92.25 93.45 i-adjp 152 145 121 83.45 79.61 81.48 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.41 74.57 75.96 Avg2. 46451 46451 45301 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12234 11770 96.21 96.32 96.26 pp 4633 4666 4569 97.92 98.62 98.27 vp 4768 4734 4558 96.28 95.60 95.94 sbar 503 490 458 93.47 91.05 92.25 adjp 384 373 322 86.33 83.85 85.07 advp 822 816 715 87.62 86.98 87.30 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.48 80.66 81.56 Avg2. 23486 23468 22537 96.03 95.96 96.00 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 105 Log-likelihood = -51005.540447 Norm (log-likelihood gradient vector) = 1744.759583 Norm (lambda vector) = 262.969666 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4580 98.18 98.86 98.52 i-np 13660 13712 13404 97.75 98.13 97.94 e-np 12220 12228 12029 98.37 98.44 98.40 o 6349 6337 6173 97.41 97.23 97.32 e-vp 4768 4734 4662 98.48 97.78 98.13 i-vp 2602 2640 2522 95.53 96.93 96.22 e-adjp 384 370 333 90.00 86.72 88.33 i-pp 52 42 36 85.71 69.23 76.60 e-advp 822 815 721 88.47 87.71 88.09 i-advp 100 87 71 81.61 71.00 75.94 e-sbar 503 489 464 94.89 92.25 93.55 i-adjp 152 143 119 83.22 78.29 80.68 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.35 74.46 75.88 Avg2. 46451 46451 45289 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11768 96.24 96.30 96.27 pp 4633 4665 4570 97.96 98.64 98.30 vp 4768 4734 4556 96.24 95.55 95.90 sbar 503 489 458 93.66 91.05 92.34 adjp 384 370 319 86.22 83.07 84.62 advp 822 815 714 87.61 86.86 87.23 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.49 80.57 81.52 Avg2. 23486 23456 22530 96.05 95.93 95.99 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 106 Log-likelihood = -50561.410221 Norm (log-likelihood gradient vector) = 1570.973766 Norm (lambda vector) = 264.100652 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4578 98.24 98.81 98.53 i-np 13660 13714 13409 97.78 98.16 97.97 e-np 12220 12230 12032 98.38 98.46 98.42 o 6349 6340 6178 97.44 97.31 97.38 e-vp 4768 4732 4662 98.52 97.78 98.15 i-vp 2602 2640 2522 95.53 96.93 96.22 e-adjp 384 372 334 89.78 86.98 88.36 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 816 722 88.48 87.83 88.16 i-advp 100 85 70 82.35 70.00 75.68 e-sbar 503 491 465 94.70 92.45 93.56 i-adjp 152 139 119 85.61 78.29 81.79 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 23 22 95.65 91.67 93.62 e-conjp 16 14 13 92.86 81.25 86.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.64 74.97 76.28 Avg2. 46451 46451 45304 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11774 96.27 96.35 96.31 pp 4633 4660 4568 98.03 98.60 98.31 vp 4768 4732 4555 96.26 95.53 95.89 sbar 503 491 459 93.48 91.25 92.35 adjp 384 372 321 86.29 83.59 84.92 advp 822 816 715 87.62 86.98 87.30 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 13 92.86 81.25 86.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.55 81.28 81.91 Avg2. 23486 23457 22538 96.08 95.96 96.02 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 107 Log-likelihood = -49615.791409 Norm (log-likelihood gradient vector) = 1632.987269 Norm (lambda vector) = 266.974022 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4632 4558 98.40 98.38 98.39 i-np 13660 13723 13407 97.70 98.15 97.92 e-np 12220 12216 12027 98.45 98.42 98.44 o 6349 6342 6179 97.43 97.32 97.38 e-vp 4768 4731 4663 98.56 97.80 98.18 i-vp 2602 2644 2528 95.61 97.16 96.38 e-adjp 384 377 336 89.12 87.50 88.30 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 817 721 88.25 87.71 87.98 i-advp 100 85 70 82.35 70.00 75.68 e-sbar 503 517 473 91.49 94.04 92.75 i-adjp 152 139 119 85.61 78.29 81.79 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.39 74.23 75.78 Avg2. 46451 46451 45290 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12216 11765 96.31 96.28 96.29 pp 4633 4632 4548 98.19 98.17 98.18 vp 4768 4731 4560 96.39 95.64 96.01 sbar 503 517 467 90.33 92.84 91.57 adjp 384 377 324 85.94 84.38 85.15 advp 822 817 713 87.27 86.74 87.00 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.14 80.82 81.48 Avg2. 23486 23445 22522 96.06 95.90 95.98 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 108 Log-likelihood = -49556.760504 Norm (log-likelihood gradient vector) = 5874.572259 Norm (lambda vector) = 269.626001 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4645 4570 98.39 98.64 98.51 i-np 13660 13729 13412 97.69 98.18 97.94 e-np 12220 12218 12029 98.45 98.44 98.45 o 6349 6335 6178 97.52 97.31 97.41 e-vp 4768 4732 4664 98.56 97.82 98.19 i-vp 2602 2641 2526 95.65 97.08 96.36 e-adjp 384 374 334 89.30 86.98 88.13 i-pp 52 41 36 87.80 69.23 77.42 e-advp 822 817 722 88.37 87.83 88.10 i-advp 100 85 70 82.35 70.00 75.68 e-sbar 503 505 472 93.47 93.84 93.65 i-adjp 152 138 118 85.51 77.63 81.38 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 23 22 95.65 91.67 93.62 e-conjp 16 14 13 92.86 81.25 86.67 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.56 75.01 76.26 Avg2. 46451 46451 45308 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12218 11771 96.34 96.33 96.33 pp 4633 4645 4560 98.17 98.42 98.30 vp 4768 4732 4560 96.37 95.64 96.00 sbar 503 505 466 92.28 92.64 92.46 adjp 384 374 322 86.10 83.85 84.96 advp 822 817 714 87.39 86.86 87.13 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 13 92.86 81.25 86.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.42 81.42 81.92 Avg2. 23486 23447 22539 96.13 95.97 96.05 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 109 Log-likelihood = -48999.315166 Norm (log-likelihood gradient vector) = 2364.448859 Norm (lambda vector) = 268.140877 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4578 98.20 98.81 98.50 i-np 13660 13722 13406 97.70 98.14 97.92 e-np 12220 12216 12025 98.44 98.40 98.42 o 6349 6337 6176 97.46 97.28 97.37 e-vp 4768 4733 4665 98.56 97.84 98.20 i-vp 2602 2642 2528 95.69 97.16 96.41 e-adjp 384 380 337 88.68 87.76 88.22 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 815 720 88.34 87.59 87.97 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 493 466 94.52 92.64 93.57 i-adjp 152 139 119 85.61 78.29 81.79 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.56 74.19 75.84 Avg2. 46451 46451 45299 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12216 11768 96.33 96.30 96.32 pp 4633 4662 4568 97.98 98.60 98.29 vp 4768 4733 4564 96.43 95.72 96.07 sbar 503 493 460 93.31 91.45 92.37 adjp 384 380 325 85.53 84.64 85.08 advp 822 815 712 87.36 86.62 86.99 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.39 80.75 81.57 Avg2. 23486 23454 22542 96.11 95.98 96.05 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 110 Log-likelihood = -47819.287967 Norm (log-likelihood gradient vector) = 1542.141119 Norm (lambda vector) = 271.908402 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4580 98.24 98.86 98.55 i-np 13660 13682 13398 97.92 98.08 98.00 e-np 12220 12234 12037 98.39 98.50 98.45 o 6349 6354 6187 97.37 97.45 97.41 e-vp 4768 4738 4667 98.50 97.88 98.19 i-vp 2602 2640 2527 95.72 97.12 96.41 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 816 721 88.36 87.71 88.03 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 494 467 94.53 92.84 93.68 i-adjp 152 139 119 85.61 78.29 81.79 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.48 74.22 75.81 Avg2. 46451 46451 45319 97.56 97.56 97.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12234 11782 96.31 96.42 96.36 pp 4633 4662 4569 98.01 98.62 98.31 vp 4768 4738 4567 96.39 95.78 96.09 sbar 503 494 461 93.32 91.65 92.48 adjp 384 379 325 85.75 84.64 85.19 advp 822 816 713 87.38 86.74 87.06 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.42 80.81 81.60 Avg2. 23486 23478 22562 96.10 96.07 96.08 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 111 Log-likelihood = -47082.233614 Norm (log-likelihood gradient vector) = 1726.343660 Norm (lambda vector) = 274.077258 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4581 98.14 98.88 98.51 i-np 13660 13690 13399 97.87 98.09 97.98 e-np 12220 12229 12038 98.44 98.51 98.47 o 6349 6348 6188 97.48 97.46 97.47 e-vp 4768 4738 4667 98.50 97.88 98.19 i-vp 2602 2644 2529 95.65 97.19 96.42 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 818 723 88.39 87.96 88.17 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 488 463 94.88 92.05 93.44 i-adjp 152 140 119 85.00 78.29 81.51 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 11 9 81.82 75.00 78.26 i-conjp 24 20 19 95.00 79.17 86.36 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.81 73.68 75.68 Avg2. 46451 46451 45321 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12229 11782 96.34 96.42 96.38 pp 4633 4668 4570 97.90 98.64 98.27 vp 4768 4738 4567 96.39 95.78 96.09 sbar 503 488 458 93.85 91.05 92.43 adjp 384 379 324 85.49 84.38 84.93 advp 822 818 715 87.41 86.98 87.20 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.37 80.12 81.23 Avg2. 23486 23474 22560 96.11 96.06 96.08 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 112 Log-likelihood = -46348.065063 Norm (log-likelihood gradient vector) = 1179.727051 Norm (lambda vector) = 275.359483 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4582 98.12 98.90 98.51 i-np 13660 13685 13404 97.95 98.13 98.04 e-np 12220 12235 12042 98.42 98.54 98.48 o 6349 6353 6193 97.48 97.54 97.51 e-vp 4768 4737 4667 98.52 97.88 98.20 i-vp 2602 2644 2528 95.61 97.16 96.38 e-adjp 384 378 337 89.15 87.76 88.45 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 812 720 88.67 87.59 88.13 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 486 461 94.86 91.65 93.23 i-adjp 152 142 121 85.21 79.61 82.31 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 12 9 75.00 75.00 75.00 i-conjp 24 20 19 95.00 79.17 86.36 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.56 73.67 75.56 Avg2. 46451 46451 45331 97.59 97.59 97.59 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12235 11791 96.37 96.49 96.43 pp 4633 4670 4572 97.90 98.68 98.29 vp 4768 4737 4566 96.39 95.76 96.08 sbar 503 486 455 93.62 90.46 92.01 adjp 384 378 323 85.45 84.11 84.78 advp 822 812 712 87.68 86.62 87.15 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.37 79.93 81.13 Avg2. 23486 23471 22562 96.13 96.07 96.10 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 113 Log-likelihood = -45611.208657 Norm (log-likelihood gradient vector) = 1202.963909 Norm (lambda vector) = 276.269145 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4582 98.16 98.90 98.53 i-np 13660 13816 13431 97.21 98.32 97.77 e-np 12220 12171 11992 98.53 98.13 98.33 o 6349 6285 6142 97.72 96.74 97.23 e-vp 4768 4734 4667 98.58 97.88 98.23 i-vp 2602 2649 2532 95.58 97.31 96.44 e-adjp 384 377 335 88.86 87.24 88.04 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 811 721 88.90 87.71 88.30 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 489 464 94.89 92.25 93.55 i-adjp 152 138 117 84.78 76.97 80.69 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.23 74.23 75.70 Avg2. 46451 46451 45262 97.44 97.44 97.44 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12171 11711 96.22 95.83 96.03 pp 4633 4668 4572 97.94 98.68 98.31 vp 4768 4734 4565 96.43 95.74 96.09 sbar 503 489 457 93.46 90.85 92.14 adjp 384 377 320 84.88 83.33 84.10 advp 822 811 713 87.92 86.74 87.32 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.38 80.46 81.41 Avg2. 23486 23404 22482 96.06 95.73 95.89 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 114 Log-likelihood = -45001.959967 Norm (log-likelihood gradient vector) = 3887.794681 Norm (lambda vector) = 278.881119 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4582 98.16 98.90 98.53 i-np 13660 13685 13389 97.84 98.02 97.93 e-np 12220 12237 12034 98.34 98.48 98.41 o 6349 6347 6181 97.38 97.35 97.37 e-vp 4768 4732 4665 98.58 97.84 98.21 i-vp 2602 2651 2532 95.51 97.31 96.40 e-adjp 384 376 332 88.30 86.46 87.37 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 815 723 88.71 87.96 88.33 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 488 464 95.08 92.25 93.64 i-adjp 152 138 116 84.06 76.32 80.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.12 74.20 75.63 Avg2. 46451 46451 45297 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12237 11765 96.14 96.28 96.21 pp 4633 4668 4572 97.94 98.68 98.31 vp 4768 4732 4562 96.41 95.68 96.04 sbar 503 488 457 93.65 90.85 92.23 adjp 384 376 317 84.31 82.55 83.42 advp 822 815 715 87.73 86.98 87.35 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.31 80.44 81.37 Avg2. 23486 23470 22532 96.00 95.94 95.97 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 115 Log-likelihood = -44453.558847 Norm (log-likelihood gradient vector) = 2384.179697 Norm (lambda vector) = 279.884840 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4582 98.20 98.90 98.55 i-np 13660 13695 13397 97.82 98.07 97.95 e-np 12220 12231 12033 98.38 98.47 98.43 o 6349 6348 6183 97.40 97.39 97.39 e-vp 4768 4729 4662 98.58 97.78 98.18 i-vp 2602 2654 2532 95.40 97.31 96.35 e-adjp 384 374 333 89.04 86.72 87.86 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 816 723 88.60 87.96 88.28 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 489 464 94.89 92.25 93.55 i-adjp 152 135 115 85.19 75.66 80.14 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.19 74.19 75.66 Avg2. 46451 46451 45303 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12231 11772 96.25 96.33 96.29 pp 4633 4666 4572 97.99 98.68 98.33 vp 4768 4729 4556 96.34 95.55 95.95 sbar 503 489 457 93.46 90.85 92.14 adjp 384 374 318 85.03 82.81 83.91 advp 822 816 715 87.62 86.98 87.30 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.36 80.46 81.40 Avg2. 23486 23459 22534 96.06 95.95 96.00 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 116 Log-likelihood = -44120.644179 Norm (log-likelihood gradient vector) = 1431.358667 Norm (lambda vector) = 280.121805 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4656 4578 98.32 98.81 98.57 i-np 13660 13717 13404 97.72 98.13 97.92 e-np 12220 12222 12027 98.40 98.42 98.41 o 6349 6344 6180 97.41 97.34 97.38 e-vp 4768 4730 4663 98.58 97.80 98.19 i-vp 2602 2657 2533 95.33 97.35 96.33 e-adjp 384 368 330 89.67 85.94 87.77 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 815 721 88.47 87.71 88.09 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 497 468 94.16 93.04 93.60 i-adjp 152 131 113 86.26 74.34 79.86 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.45 74.52 75.96 Avg2. 46451 46451 45297 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12222 11766 96.27 96.28 96.28 pp 4633 4656 4568 98.11 98.60 98.35 vp 4768 4730 4555 96.30 95.53 95.91 sbar 503 497 462 92.96 91.85 92.40 adjp 384 368 316 85.87 82.29 84.04 advp 822 815 713 87.48 86.74 87.11 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.40 80.47 81.42 Avg2. 23486 23442 22524 96.08 95.90 95.99 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 117 Log-likelihood = -43577.193757 Norm (log-likelihood gradient vector) = 1099.693517 Norm (lambda vector) = 282.021504 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4656 4579 98.35 98.83 98.59 i-np 13660 13754 13417 97.55 98.22 97.88 e-np 12220 12210 12016 98.41 98.33 98.37 o 6349 6331 6171 97.47 97.20 97.33 e-vp 4768 4726 4660 98.60 97.73 98.17 i-vp 2602 2657 2534 95.37 97.39 96.37 e-adjp 384 367 330 89.92 85.94 87.88 i-pp 52 42 35 83.33 67.31 74.47 e-advp 822 810 718 88.64 87.35 87.99 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 494 466 94.33 92.64 93.48 i-adjp 152 130 113 86.92 74.34 80.14 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.30 74.48 75.86 Avg2. 46451 46451 45284 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12210 11750 96.23 96.15 96.19 pp 4633 4656 4567 98.09 98.58 98.33 vp 4768 4726 4555 96.38 95.53 95.96 sbar 503 494 460 93.12 91.45 92.28 adjp 384 367 316 86.10 82.29 84.15 advp 822 810 710 87.65 86.37 87.01 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.45 80.38 81.40 Avg2. 23486 23417 22502 96.09 95.81 95.95 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 118 Log-likelihood = -43071.415682 Norm (log-likelihood gradient vector) = 1284.277277 Norm (lambda vector) = 284.142739 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4656 4578 98.32 98.81 98.57 i-np 13660 13759 13424 97.57 98.27 97.92 e-np 12220 12206 12018 98.46 98.35 98.40 o 6349 6326 6171 97.55 97.20 97.37 e-vp 4768 4730 4663 98.58 97.80 98.19 i-vp 2602 2651 2532 95.51 97.31 96.40 e-adjp 384 367 330 89.92 85.94 87.88 i-pp 52 42 35 83.33 67.31 74.47 e-advp 822 814 720 88.45 87.59 88.02 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 496 467 94.15 92.84 93.49 i-adjp 152 129 112 86.82 73.68 79.72 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.39 74.88 76.12 Avg2. 46451 46451 45296 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12206 11758 96.33 96.22 96.27 pp 4633 4656 4566 98.07 98.55 98.31 vp 4768 4730 4558 96.36 95.60 95.98 sbar 503 496 462 93.15 91.85 92.49 adjp 384 367 315 85.83 82.03 83.89 advp 822 814 712 87.47 86.62 87.04 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.42 80.43 81.41 Avg2. 23486 23423 22515 96.12 95.87 95.99 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 119 Log-likelihood = -42253.545726 Norm (log-likelihood gradient vector) = 1614.457411 Norm (lambda vector) = 288.953101 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4673 4585 98.12 98.96 98.54 i-np 13660 13697 13389 97.75 98.02 97.88 e-np 12220 12237 12031 98.32 98.45 98.38 o 6349 6354 6182 97.29 97.37 97.33 e-vp 4768 4736 4666 98.52 97.86 98.19 i-vp 2602 2644 2530 95.69 97.23 96.45 e-adjp 384 370 331 89.46 86.20 87.80 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 816 721 88.36 87.71 88.03 i-advp 100 85 70 82.35 70.00 75.68 e-sbar 503 478 458 95.82 91.05 93.37 i-adjp 152 129 111 86.05 73.03 79.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.44 74.79 76.09 Avg2. 46451 46451 45285 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12237 11761 96.11 96.24 96.18 pp 4633 4673 4574 97.88 98.73 98.30 vp 4768 4736 4565 96.39 95.74 96.06 sbar 503 478 453 94.77 90.06 92.35 adjp 384 370 315 85.14 82.03 83.55 advp 822 816 713 87.38 86.74 87.06 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.46 80.30 81.36 Avg2. 23486 23464 22525 96.00 95.91 95.95 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 120 Log-likelihood = -41569.883588 Norm (log-likelihood gradient vector) = 3057.089653 Norm (lambda vector) = 294.260997 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4583 98.18 98.92 98.55 i-np 13660 13717 13411 97.77 98.18 97.97 e-np 12220 12226 12032 98.41 98.46 98.44 o 6349 6344 6181 97.43 97.35 97.39 e-vp 4768 4737 4667 98.52 97.88 98.20 i-vp 2602 2645 2530 95.65 97.23 96.44 e-adjp 384 371 332 89.49 86.46 87.95 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 814 721 88.57 87.71 88.14 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 482 461 95.64 91.65 93.60 i-adjp 152 131 113 86.26 74.34 79.86 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.52 74.90 76.19 Avg2. 46451 46451 45312 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12226 11770 96.27 96.32 96.29 pp 4633 4668 4572 97.94 98.68 98.31 vp 4768 4737 4566 96.39 95.76 96.08 sbar 503 482 456 94.61 90.66 92.59 adjp 384 371 316 85.18 82.29 83.71 advp 822 814 713 87.59 86.74 87.16 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.49 80.39 81.43 Avg2. 23486 23452 22537 96.10 95.96 96.03 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 121 Log-likelihood = -41107.277898 Norm (log-likelihood gradient vector) = 1228.157390 Norm (lambda vector) = 293.289860 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4580 98.20 98.86 98.53 i-np 13660 13713 13407 97.77 98.15 97.96 e-np 12220 12222 12033 98.45 98.47 98.46 o 6349 6343 6184 97.49 97.40 97.45 e-vp 4768 4738 4667 98.50 97.88 98.19 i-vp 2602 2650 2532 95.55 97.31 96.42 e-adjp 384 371 332 89.49 86.46 87.95 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 814 721 88.57 87.71 88.14 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 489 463 94.68 92.05 93.35 i-adjp 152 131 113 86.26 74.34 79.86 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.47 74.93 76.18 Avg2. 46451 46451 45313 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12222 11772 96.32 96.33 96.33 pp 4633 4664 4569 97.96 98.62 98.29 vp 4768 4738 4566 96.37 95.76 96.07 sbar 503 489 458 93.66 91.05 92.34 adjp 384 371 316 85.18 82.29 83.71 advp 822 814 713 87.59 86.74 87.16 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.40 80.42 81.40 Avg2. 23486 23452 22538 96.10 95.96 96.03 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 122 Log-likelihood = -40836.925296 Norm (log-likelihood gradient vector) = 885.720908 Norm (lambda vector) = 293.242979 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4579 98.20 98.83 98.52 i-np 13660 13720 13409 97.73 98.16 97.95 e-np 12220 12217 12030 98.47 98.45 98.46 o 6349 6336 6177 97.49 97.29 97.39 e-vp 4768 4739 4668 98.50 97.90 98.20 i-vp 2602 2653 2532 95.44 97.31 96.37 e-adjp 384 370 334 90.27 86.98 88.59 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 814 721 88.57 87.71 88.14 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 488 463 94.88 92.05 93.44 i-adjp 152 136 114 83.82 75.00 79.17 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.55 75.02 76.26 Avg2. 46451 46451 45309 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12217 11768 96.32 96.30 96.31 pp 4633 4663 4569 97.98 98.62 98.30 vp 4768 4739 4566 96.35 95.76 96.06 sbar 503 488 458 93.85 91.05 92.43 adjp 384 370 317 85.68 82.55 84.08 advp 822 814 713 87.59 86.74 87.16 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.48 80.52 81.49 Avg2. 23486 23446 22536 96.12 95.96 96.04 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 123 Log-likelihood = -40475.058696 Norm (log-likelihood gradient vector) = 1114.104070 Norm (lambda vector) = 294.185717 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4578 98.20 98.81 98.50 i-np 13660 13565 13342 98.36 97.67 98.01 e-np 12220 12289 12066 98.19 98.74 98.46 o 6349 6413 6218 96.96 97.94 97.45 e-vp 4768 4748 4674 98.44 98.03 98.23 i-vp 2602 2644 2530 95.69 97.23 96.45 e-adjp 384 374 335 89.57 87.24 88.39 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 816 721 88.36 87.71 88.03 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 492 464 94.31 92.25 93.27 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 74.59 75.98 Avg2. 46451 46451 45323 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12289 11816 96.15 96.69 96.42 pp 4633 4662 4567 97.96 98.58 98.27 vp 4768 4748 4574 96.34 95.93 96.13 sbar 503 492 458 93.09 91.05 92.06 adjp 384 374 319 85.29 83.07 84.17 advp 822 816 713 87.38 86.74 87.06 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.25 80.63 81.43 Avg2. 23486 23537 22592 95.99 96.19 96.09 Current max chunk-based F1: 96.13 (iteration 88) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 124 Log-likelihood = -40212.196406 Norm (log-likelihood gradient vector) = 5772.053100 Norm (lambda vector) = 298.033388 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 320 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4581 98.20 98.88 98.54 i-np 13660 13638 13389 98.17 98.02 98.10 e-np 12220 12257 12055 98.35 98.65 98.50 o 6349 6377 6204 97.29 97.72 97.50 e-vp 4768 4742 4671 98.50 97.97 98.23 i-vp 2602 2648 2532 95.62 97.31 96.46 e-adjp 384 373 335 89.81 87.24 88.51 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 817 722 88.37 87.83 88.10 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 488 463 94.88 92.05 93.44 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.55 74.59 76.04 Avg2. 46451 46451 45347 97.62 97.62 97.62 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12257 11810 96.35 96.64 96.50 pp 4633 4665 4570 97.96 98.64 98.30 vp 4768 4742 4571 96.39 95.87 96.13 sbar 503 488 457 93.65 90.85 92.23 adjp 384 373 319 85.52 83.07 84.28 advp 822 817 714 87.39 86.86 87.13 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.43 80.61 81.51 Avg2. 23486 23497 22586 96.12 96.17 96.15 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 125 Log-likelihood = -40128.316594 Norm (log-likelihood gradient vector) = 3102.743635 Norm (lambda vector) = 296.080400 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4577 98.20 98.79 98.49 i-np 13660 13673 13402 98.02 98.11 98.06 e-np 12220 12238 12044 98.41 98.56 98.49 o 6349 6360 6195 97.41 97.57 97.49 e-vp 4768 4746 4674 98.48 98.03 98.26 i-vp 2602 2645 2530 95.65 97.23 96.44 e-adjp 384 373 335 89.81 87.24 88.51 i-pp 52 39 34 87.18 65.38 74.73 e-advp 822 815 720 88.34 87.59 87.97 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 492 464 94.31 92.25 93.27 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.44 74.57 75.98 Avg2. 46451 46451 45336 97.60 97.60 97.60 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12238 11796 96.39 96.53 96.46 pp 4633 4661 4566 97.96 98.55 98.26 vp 4768 4746 4573 96.35 95.91 96.13 sbar 503 492 458 93.09 91.05 92.06 adjp 384 373 319 85.52 83.07 84.28 advp 822 815 712 87.36 86.62 86.99 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.30 80.59 81.44 Avg2. 23486 23481 22569 96.12 96.10 96.11 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 126 Log-likelihood = -39467.084010 Norm (log-likelihood gradient vector) = 1802.887323 Norm (lambda vector) = 298.505535 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4577 98.20 98.79 98.49 i-np 13660 13700 13414 97.91 98.20 98.06 e-np 12220 12230 12039 98.44 98.52 98.48 o 6349 6349 6189 97.48 97.48 97.48 e-vp 4768 4746 4675 98.50 98.05 98.28 i-vp 2602 2643 2530 95.72 97.23 96.47 e-adjp 384 369 331 89.70 86.20 87.92 i-pp 52 39 33 84.62 63.46 72.53 e-advp 822 815 719 88.22 87.47 87.84 i-advp 100 83 70 84.34 70.00 76.50 e-sbar 503 491 464 94.50 92.25 93.36 i-adjp 152 131 114 87.02 75.00 80.57 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.85 74.38 75.60 Avg2. 46451 46451 45331 97.59 97.59 97.59 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11789 96.39 96.47 96.43 pp 4633 4661 4566 97.96 98.55 98.26 vp 4768 4746 4575 96.40 95.95 96.17 sbar 503 491 458 93.28 91.05 92.15 adjp 384 369 316 85.64 82.29 83.93 advp 822 815 712 87.36 86.62 86.99 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.67 80.51 81.09 Avg2. 23486 23469 22561 96.13 96.06 96.10 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 127 Log-likelihood = -38876.706588 Norm (log-likelihood gradient vector) = 1014.280452 Norm (lambda vector) = 300.939511 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4658 4578 98.28 98.81 98.55 i-np 13660 13723 13423 97.81 98.27 98.04 e-np 12220 12222 12033 98.45 98.47 98.46 o 6349 6342 6182 97.48 97.37 97.42 e-vp 4768 4747 4678 98.55 98.11 98.33 i-vp 2602 2637 2526 95.79 97.08 96.43 e-adjp 384 366 330 90.16 85.94 88.00 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 817 721 88.25 87.71 87.98 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 491 465 94.70 92.45 93.56 i-adjp 152 131 114 87.02 75.00 80.57 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.42 74.58 75.97 Avg2. 46451 46451 45331 97.59 97.59 97.59 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12222 11782 96.40 96.42 96.41 pp 4633 4658 4568 98.07 98.60 98.33 vp 4768 4747 4576 96.40 95.97 96.18 sbar 503 491 459 93.48 91.25 92.35 adjp 384 366 315 86.07 82.03 84.00 advp 822 817 714 87.39 86.86 87.13 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.41 80.53 81.46 Avg2. 23486 23457 22559 96.17 96.05 96.11 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 128 Log-likelihood = -38335.688857 Norm (log-likelihood gradient vector) = 1086.718680 Norm (lambda vector) = 303.101796 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4580 98.28 98.86 98.57 i-np 13660 13728 13422 97.77 98.26 98.01 e-np 12220 12220 12031 98.45 98.45 98.45 o 6349 6342 6179 97.43 97.32 97.38 e-vp 4768 4744 4676 98.57 98.07 98.32 i-vp 2602 2634 2524 95.82 97.00 96.41 e-adjp 384 367 331 90.19 86.20 88.15 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 821 721 87.82 87.71 87.77 i-advp 100 84 70 83.33 70.00 76.09 e-sbar 503 489 464 94.89 92.25 93.55 i-adjp 152 131 114 87.02 75.00 80.57 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.36 74.05 75.67 Avg2. 46451 46451 45321 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12220 11775 96.36 96.36 96.36 pp 4633 4660 4570 98.07 98.64 98.35 vp 4768 4744 4573 96.40 95.91 96.15 sbar 503 489 458 93.66 91.05 92.34 adjp 384 367 316 86.10 82.29 84.15 advp 822 821 714 86.97 86.86 86.91 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.32 79.91 81.10 Avg2. 23486 23456 22550 96.14 96.01 96.08 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 129 Log-likelihood = -37768.493306 Norm (log-likelihood gradient vector) = 1228.688910 Norm (lambda vector) = 305.161602 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4654 4577 98.35 98.79 98.57 i-np 13660 13730 13428 97.80 98.30 98.05 e-np 12220 12213 12029 98.49 98.44 98.47 o 6349 6349 6187 97.45 97.45 97.45 e-vp 4768 4739 4670 98.54 97.94 98.24 i-vp 2602 2635 2522 95.71 96.93 96.31 e-adjp 384 367 330 89.92 85.94 87.88 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 823 722 87.73 87.83 87.78 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 495 467 94.34 92.84 93.59 i-adjp 152 130 113 86.92 74.34 80.14 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.38 74.09 75.70 Avg2. 46451 46451 45325 97.58 97.58 97.58 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12213 11780 96.45 96.40 96.43 pp 4633 4654 4567 98.13 98.58 98.35 vp 4768 4739 4563 96.29 95.70 95.99 sbar 503 495 461 93.13 91.65 92.38 adjp 384 367 315 85.83 82.03 83.89 advp 822 823 715 86.88 86.98 86.93 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.24 79.93 81.07 Avg2. 23486 23446 22545 96.16 95.99 96.08 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 130 Log-likelihood = -36680.661848 Norm (log-likelihood gradient vector) = 2132.984776 Norm (lambda vector) = 309.406435 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4581 98.18 98.88 98.53 i-np 13660 13721 13411 97.74 98.18 97.96 e-np 12220 12211 12024 98.47 98.40 98.43 o 6349 6347 6182 97.40 97.37 97.39 e-vp 4768 4739 4670 98.54 97.94 98.24 i-vp 2602 2637 2524 95.71 97.00 96.35 e-adjp 384 372 332 89.25 86.46 87.83 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 826 724 87.65 88.08 87.86 i-advp 100 87 71 81.61 71.00 75.94 e-sbar 503 483 460 95.24 91.45 93.31 i-adjp 152 131 114 87.02 75.00 80.57 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.22 74.08 75.62 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12211 11764 96.34 96.27 96.30 pp 4633 4666 4571 97.96 98.66 98.31 vp 4768 4739 4565 96.33 95.74 96.03 sbar 503 483 454 94.00 90.26 92.09 adjp 384 372 318 85.48 82.81 84.13 advp 822 826 717 86.80 87.23 87.01 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.26 79.89 81.06 Avg2. 23486 23452 22533 96.08 95.94 96.01 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 131 Log-likelihood = -35832.333619 Norm (log-likelihood gradient vector) = 1617.782508 Norm (lambda vector) = 313.745316 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4580 98.24 98.86 98.55 i-np 13660 13696 13406 97.88 98.14 98.01 e-np 12220 12219 12032 98.47 98.46 98.47 o 6349 6357 6190 97.37 97.50 97.43 e-vp 4768 4739 4671 98.57 97.97 98.26 i-vp 2602 2638 2526 95.75 97.08 96.41 e-adjp 384 373 334 89.54 86.98 88.24 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 825 723 87.64 87.96 87.80 i-advp 100 87 71 81.61 71.00 75.94 e-sbar 503 490 464 94.69 92.25 93.45 i-adjp 152 134 117 87.31 76.97 81.82 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.24 74.25 75.72 Avg2. 46451 46451 45323 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12219 11775 96.37 96.36 96.36 pp 4633 4662 4570 98.03 98.64 98.33 vp 4768 4739 4567 96.37 95.78 96.08 sbar 503 490 458 93.47 91.05 92.25 adjp 384 373 320 85.79 83.33 84.54 advp 822 825 716 86.79 87.10 86.95 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.25 80.02 81.12 Avg2. 23486 23463 22550 96.11 96.01 96.06 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 132 Log-likelihood = -35555.827013 Norm (log-likelihood gradient vector) = 969.471339 Norm (lambda vector) = 313.404387 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4579 98.26 98.83 98.55 i-np 13660 13686 13404 97.94 98.13 98.03 e-np 12220 12225 12036 98.45 98.49 98.47 o 6349 6355 6190 97.40 97.50 97.45 e-vp 4768 4740 4672 98.57 97.99 98.28 i-vp 2602 2643 2528 95.65 97.16 96.40 e-adjp 384 375 335 89.33 87.24 88.27 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 824 722 87.62 87.83 87.73 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 493 466 94.52 92.64 93.57 i-adjp 152 134 116 86.57 76.32 81.12 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.33 74.25 75.76 Avg2. 46451 46451 45328 97.58 97.58 97.58 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12225 11783 96.38 96.42 96.40 pp 4633 4660 4569 98.05 98.62 98.33 vp 4768 4740 4567 96.35 95.78 96.07 sbar 503 493 460 93.31 91.45 92.37 adjp 384 375 319 85.07 83.07 84.06 advp 822 824 715 86.77 86.98 86.88 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.16 80.03 81.08 Avg2. 23486 23472 22557 96.10 96.04 96.07 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 133 Log-likelihood = -35216.188020 Norm (log-likelihood gradient vector) = 1046.537146 Norm (lambda vector) = 313.465099 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4656 4577 98.30 98.79 98.55 i-np 13660 13701 13403 97.82 98.12 97.97 e-np 12220 12219 12033 98.48 98.47 98.47 o 6349 6348 6186 97.45 97.43 97.44 e-vp 4768 4742 4672 98.52 97.99 98.25 i-vp 2602 2642 2528 95.69 97.16 96.41 e-adjp 384 374 334 89.30 86.98 88.13 i-pp 52 40 35 87.50 67.31 76.09 e-advp 822 826 723 87.53 87.96 87.74 i-advp 100 87 71 81.61 71.00 75.94 e-sbar 503 494 467 94.53 92.84 93.68 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.27 74.18 75.70 Avg2. 46451 46451 45317 97.56 97.56 97.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12219 11778 96.39 96.38 96.39 pp 4633 4656 4567 98.09 98.58 98.33 vp 4768 4742 4568 96.33 95.81 96.07 sbar 503 494 461 93.32 91.65 92.48 adjp 384 374 318 85.03 82.81 83.91 advp 822 826 715 86.56 86.98 86.77 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.14 80.02 81.06 Avg2. 23486 23466 22551 96.10 96.02 96.06 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 134 Log-likelihood = -34889.282446 Norm (log-likelihood gradient vector) = 1003.011325 Norm (lambda vector) = 314.375850 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4655 4580 98.39 98.86 98.62 i-np 13660 13363 13197 98.76 96.61 97.67 e-np 12220 12395 12096 97.59 98.99 98.28 o 6349 6518 6249 95.87 98.42 97.13 e-vp 4768 4740 4665 98.42 97.84 98.13 i-vp 2602 2639 2524 95.64 97.00 96.32 e-adjp 384 371 332 89.49 86.46 87.95 i-pp 52 40 34 85.00 65.38 73.91 e-advp 822 820 722 88.05 87.83 87.94 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 498 469 94.18 93.24 93.71 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.06 74.06 75.05 Avg2. 46451 46451 45227 97.36 97.36 97.36 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12395 11814 95.31 96.68 95.99 pp 4633 4655 4569 98.15 98.62 98.39 vp 4768 4740 4561 96.22 95.66 95.94 sbar 503 498 463 92.97 92.05 92.51 adjp 384 371 316 85.18 82.29 83.71 advp 822 820 714 87.07 86.86 86.97 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.75 80.01 80.38 Avg2. 23486 23636 22581 95.54 96.15 95.84 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 135 Log-likelihood = -38629.162924 Norm (log-likelihood gradient vector) = 15274.931554 Norm (lambda vector) = 322.102228 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4658 4579 98.30 98.83 98.57 i-np 13660 13641 13378 98.07 97.94 98.00 e-np 12220 12247 12046 98.36 98.58 98.47 o 6349 6380 6201 97.19 97.67 97.43 e-vp 4768 4740 4672 98.57 97.99 98.28 i-vp 2602 2642 2528 95.69 97.16 96.41 e-adjp 384 374 335 89.57 87.24 88.39 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 825 723 87.64 87.96 87.80 i-advp 100 87 71 81.61 71.00 75.94 e-sbar 503 494 467 94.53 92.84 93.68 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 133 125 93.98 99.21 96.53 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.18 74.21 75.66 Avg2. 46451 46451 45323 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12247 11795 96.31 96.52 96.42 pp 4633 4658 4569 98.09 98.62 98.35 vp 4768 4740 4568 96.37 95.81 96.09 sbar 503 494 461 93.32 91.65 92.48 adjp 384 374 319 85.29 83.07 84.17 advp 822 825 715 86.67 86.98 86.82 prt 126 133 125 93.98 99.21 96.53 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.17 80.06 81.10 Avg2. 23486 23493 22571 96.08 96.10 96.09 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 136 Log-likelihood = -34726.543518 Norm (log-likelihood gradient vector) = 2517.116821 Norm (lambda vector) = 315.675264 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4579 98.33 98.83 98.58 i-np 13660 13685 13401 97.92 98.10 98.01 e-np 12220 12232 12038 98.41 98.51 98.46 o 6349 6362 6192 97.33 97.53 97.43 e-vp 4768 4738 4667 98.50 97.88 98.19 i-vp 2602 2640 2525 95.64 97.04 96.34 e-adjp 384 372 333 89.52 86.72 88.10 i-pp 52 41 35 85.37 67.31 75.27 e-advp 822 821 722 87.94 87.83 87.89 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 494 467 94.53 92.84 93.68 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.70 74.16 75.41 Avg2. 46451 46451 45318 97.56 97.56 97.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12232 11785 96.35 96.44 96.39 pp 4633 4657 4569 98.11 98.62 98.36 vp 4768 4738 4562 96.29 95.68 95.98 sbar 503 494 461 93.32 91.65 92.48 adjp 384 372 317 85.22 82.55 83.86 advp 822 821 714 86.97 86.86 86.91 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.56 79.98 80.76 Avg2. 23486 23469 22552 96.09 96.02 96.06 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 137 Log-likelihood = -34149.355879 Norm (log-likelihood gradient vector) = 1233.043446 Norm (lambda vector) = 318.364306 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4580 98.35 98.86 98.60 i-np 13660 13724 13412 97.73 98.18 97.96 e-np 12220 12212 12027 98.49 98.42 98.45 o 6349 6346 6185 97.46 97.42 97.44 e-vp 4768 4738 4667 98.50 97.88 98.19 i-vp 2602 2641 2525 95.61 97.04 96.32 e-adjp 384 371 333 89.76 86.72 88.21 i-pp 52 42 35 83.33 67.31 74.47 e-advp 822 818 721 88.14 87.71 87.93 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 494 467 94.53 92.84 93.68 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.67 74.15 75.39 Avg2. 46451 46451 45311 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12212 11771 96.39 96.33 96.36 pp 4633 4657 4569 98.11 98.62 98.36 vp 4768 4738 4561 96.26 95.66 95.96 sbar 503 494 461 93.32 91.65 92.48 adjp 384 371 317 85.44 82.55 83.97 advp 822 818 713 87.16 86.74 86.95 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.60 79.95 80.77 Avg2. 23486 23445 22536 96.12 95.96 96.04 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 138 Log-likelihood = -33836.832284 Norm (log-likelihood gradient vector) = 829.966270 Norm (lambda vector) = 319.975237 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4581 98.33 98.88 98.60 i-np 13660 13742 13417 97.63 98.22 97.93 e-np 12220 12205 12021 98.49 98.37 98.43 o 6349 6336 6178 97.51 97.31 97.41 e-vp 4768 4739 4666 98.46 97.86 98.16 i-vp 2602 2640 2525 95.64 97.04 96.34 e-adjp 384 370 332 89.73 86.46 88.06 i-pp 52 42 35 83.33 67.31 74.47 e-advp 822 817 720 88.13 87.59 87.86 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 493 466 94.52 92.64 93.57 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.67 74.12 75.37 Avg2. 46451 46451 45300 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12205 11763 96.38 96.26 96.32 pp 4633 4659 4570 98.09 98.64 98.36 vp 4768 4739 4560 96.22 95.64 95.93 sbar 503 493 460 93.31 91.45 92.37 adjp 384 370 316 85.41 82.29 83.82 advp 822 817 712 87.15 86.62 86.88 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.59 79.89 80.73 Avg2. 23486 23438 22525 96.10 95.91 96.01 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 139 Log-likelihood = -33347.851928 Norm (log-likelihood gradient vector) = 1015.336016 Norm (lambda vector) = 322.908205 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4582 98.31 98.90 98.60 i-np 13660 13740 13414 97.63 98.20 97.91 e-np 12220 12207 12021 98.48 98.37 98.42 o 6349 6340 6178 97.44 97.31 97.38 e-vp 4768 4737 4665 98.48 97.84 98.16 i-vp 2602 2633 2519 95.67 96.81 96.24 e-adjp 384 373 334 89.54 86.98 88.24 i-pp 52 42 35 83.33 67.31 74.47 e-advp 822 817 720 88.13 87.59 87.86 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 490 465 94.90 92.45 93.66 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.70 74.64 75.66 Avg2. 46451 46451 45294 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12207 11759 96.33 96.23 96.28 pp 4633 4661 4571 98.07 98.66 98.36 vp 4768 4737 4558 96.22 95.60 95.91 sbar 503 490 459 93.67 91.25 92.45 adjp 384 373 318 85.25 82.81 84.02 advp 822 817 712 87.15 86.62 86.88 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.71 80.54 81.12 Avg2. 23486 23441 22522 96.08 95.90 95.99 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 140 Log-likelihood = -32889.387838 Norm (log-likelihood gradient vector) = 1194.265974 Norm (lambda vector) = 325.713540 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4650 4574 98.37 98.73 98.55 i-np 13660 13699 13406 97.86 98.14 98.00 e-np 12220 12227 12032 98.41 98.46 98.43 o 6349 6362 6189 97.28 97.48 97.38 e-vp 4768 4735 4664 98.50 97.82 98.16 i-vp 2602 2634 2521 95.71 96.89 96.29 e-adjp 384 376 335 89.10 87.24 88.16 i-pp 52 41 34 82.93 65.38 73.12 e-advp 822 820 723 88.17 87.96 88.06 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 497 467 93.96 92.84 93.40 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.70 74.63 75.65 Avg2. 46451 46451 45307 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12227 11778 96.33 96.38 96.36 pp 4633 4650 4563 98.13 98.49 98.31 vp 4768 4735 4559 96.28 95.62 95.95 sbar 503 497 461 92.76 91.65 92.20 adjp 384 376 318 84.57 82.81 83.68 advp 822 820 716 87.32 87.10 87.21 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.58 80.63 81.10 Avg2. 23486 23461 22540 96.07 95.97 96.02 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 141 Log-likelihood = -32242.286168 Norm (log-likelihood gradient vector) = 2298.452626 Norm (lambda vector) = 332.275977 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4583 98.22 98.92 98.57 i-np 13660 13728 13417 97.73 98.22 97.98 e-np 12220 12207 12024 98.50 98.40 98.45 o 6349 6341 6177 97.41 97.29 97.35 e-vp 4768 4736 4664 98.48 97.82 98.15 i-vp 2602 2637 2521 95.60 96.89 96.24 e-adjp 384 375 336 89.60 87.50 88.54 i-pp 52 43 35 81.40 67.31 73.68 e-advp 822 817 721 88.25 87.71 87.98 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 489 463 94.68 92.05 93.35 i-adjp 152 136 115 84.56 75.66 79.86 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.09 74.69 75.87 Avg2. 46451 46451 45303 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12207 11769 96.41 96.31 96.36 pp 4633 4666 4572 97.99 98.68 98.33 vp 4768 4736 4559 96.26 95.62 95.94 sbar 503 489 457 93.46 90.85 92.14 adjp 384 375 318 84.80 82.81 83.79 advp 822 817 714 87.39 86.86 87.13 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.33 80.53 81.42 Avg2. 23486 23445 22534 96.11 95.95 96.03 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 142 Log-likelihood = -31745.008200 Norm (log-likelihood gradient vector) = 916.068445 Norm (lambda vector) = 333.922314 Log-likelihood and gradient computational time: 322 seconds Training iteration elapsed: 323 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4584 98.24 98.94 98.59 i-np 13660 13719 13415 97.78 98.21 97.99 e-np 12220 12210 12026 98.49 98.41 98.45 o 6349 6345 6179 97.38 97.32 97.35 e-vp 4768 4738 4665 98.46 97.84 98.15 i-vp 2602 2638 2521 95.56 96.89 96.22 e-adjp 384 375 336 89.60 87.50 88.54 i-pp 52 43 35 81.40 67.31 73.68 e-advp 822 815 719 88.22 87.47 87.84 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 489 463 94.68 92.05 93.35 i-adjp 152 137 115 83.94 75.66 79.58 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.06 74.68 75.85 Avg2. 46451 46451 45305 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12210 11773 96.42 96.34 96.38 pp 4633 4666 4573 98.01 98.70 98.35 vp 4768 4738 4559 96.22 95.62 95.92 sbar 503 489 457 93.46 90.85 92.14 adjp 384 375 317 84.53 82.55 83.53 advp 822 815 712 87.36 86.62 86.99 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.30 80.49 81.38 Avg2. 23486 23448 22536 96.11 95.96 96.03 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 357 seconds Iteration: 143 Log-likelihood = -31527.980226 Norm (log-likelihood gradient vector) = 758.050562 Norm (lambda vector) = 334.071029 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4584 98.24 98.94 98.59 i-np 13660 13691 13406 97.92 98.14 98.03 e-np 12220 12226 12035 98.44 98.49 98.46 o 6349 6360 6190 97.33 97.50 97.41 e-vp 4768 4741 4669 98.48 97.92 98.20 i-vp 2602 2636 2522 95.68 96.93 96.30 e-adjp 384 374 336 89.84 87.50 88.65 i-pp 52 43 35 81.40 67.31 73.68 e-advp 822 814 718 88.21 87.35 87.78 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 488 463 94.88 92.05 93.44 i-adjp 152 136 115 84.56 75.66 79.86 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.12 74.69 75.89 Avg2. 46451 46451 45320 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12226 11786 96.40 96.45 96.42 pp 4633 4666 4573 98.01 98.70 98.35 vp 4768 4741 4566 96.31 95.76 96.04 sbar 503 488 457 93.65 90.85 92.23 adjp 384 374 317 84.76 82.55 83.64 advp 822 814 711 87.35 86.50 86.92 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.35 80.50 81.41 Avg2. 23486 23464 22555 96.13 96.04 96.08 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 144 Log-likelihood = -31111.557000 Norm (log-likelihood gradient vector) = 966.897090 Norm (lambda vector) = 335.740609 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4580 98.28 98.86 98.57 i-np 13660 13731 13423 97.76 98.27 98.01 e-np 12220 12210 12026 98.49 98.41 98.45 o 6349 6339 6177 97.44 97.29 97.37 e-vp 4768 4741 4670 98.50 97.94 98.22 i-vp 2602 2634 2521 95.71 96.89 96.29 e-adjp 384 375 337 89.87 87.76 88.80 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 811 718 88.53 87.35 87.94 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 493 466 94.52 92.64 93.57 i-adjp 152 136 115 84.56 75.66 79.86 e-prt 126 132 125 94.70 99.21 96.90 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.66 75.04 75.84 Avg2. 46451 46451 45315 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12210 11776 96.45 96.37 96.41 pp 4633 4660 4569 98.05 98.62 98.33 vp 4768 4741 4566 96.31 95.76 96.04 sbar 503 493 461 93.51 91.65 92.57 adjp 384 375 318 84.80 82.81 83.79 advp 822 811 711 87.67 86.50 87.08 prt 126 132 125 94.70 99.21 96.90 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.72 80.59 81.15 Avg2. 23486 23446 22546 96.16 96.00 96.08 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 145 Log-likelihood = -30666.394368 Norm (log-likelihood gradient vector) = 1551.927982 Norm (lambda vector) = 338.736049 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4578 98.26 98.81 98.54 i-np 13660 13683 13401 97.94 98.10 98.02 e-np 12220 12232 12039 98.42 98.52 98.47 o 6349 6361 6187 97.26 97.45 97.36 e-vp 4768 4744 4670 98.44 97.94 98.19 i-vp 2602 2632 2521 95.78 96.89 96.33 e-adjp 384 373 334 89.54 86.98 88.24 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 815 721 88.47 87.71 88.09 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 496 466 93.95 92.64 93.29 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.64 75.03 75.83 Avg2. 46451 46451 45314 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12232 11793 96.41 96.51 96.46 pp 4633 4659 4567 98.03 98.58 98.30 vp 4768 4744 4569 96.31 95.83 96.07 sbar 503 496 461 92.94 91.65 92.29 adjp 384 373 316 84.72 82.29 83.49 advp 822 815 714 87.61 86.86 87.23 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.57 80.51 81.04 Avg2. 23486 23475 22564 96.12 96.07 96.10 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 146 Log-likelihood = -30190.730323 Norm (log-likelihood gradient vector) = 1829.278626 Norm (lambda vector) = 342.556549 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4578 98.22 98.81 98.52 i-np 13660 13684 13401 97.93 98.10 98.02 e-np 12220 12232 12039 98.42 98.52 98.47 o 6349 6361 6188 97.28 97.46 97.37 e-vp 4768 4741 4667 98.44 97.88 98.16 i-vp 2602 2637 2522 95.64 96.93 96.28 e-adjp 384 373 334 89.54 86.98 88.24 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 815 720 88.34 87.59 87.97 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 494 464 93.93 92.25 93.08 i-adjp 152 131 115 87.79 75.66 81.27 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.72 75.01 75.86 Avg2. 46451 46451 45310 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12232 11791 96.39 96.49 96.44 pp 4633 4661 4567 97.98 98.58 98.28 vp 4768 4741 4563 96.25 95.70 95.97 sbar 503 494 459 92.91 91.25 92.08 adjp 384 373 317 84.99 82.55 83.75 advp 822 815 713 87.48 86.74 87.11 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.57 80.47 81.02 Avg2. 23486 23472 22554 96.09 96.03 96.06 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 147 Log-likelihood = -29724.909168 Norm (log-likelihood gradient vector) = 936.047327 Norm (lambda vector) = 344.977761 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4578 98.26 98.81 98.54 i-np 13660 13695 13404 97.88 98.13 98.00 e-np 12220 12225 12035 98.45 98.49 98.47 o 6349 6351 6183 97.35 97.39 97.37 e-vp 4768 4743 4668 98.42 97.90 98.16 i-vp 2602 2638 2521 95.56 96.89 96.22 e-adjp 384 371 333 89.76 86.72 88.21 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 820 722 88.05 87.83 87.94 i-advp 100 88 72 81.82 72.00 76.60 e-sbar 503 495 465 93.94 92.45 93.19 i-adjp 152 130 114 87.69 75.00 80.85 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.62 74.98 75.79 Avg2. 46451 46451 45305 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12225 11783 96.38 96.42 96.40 pp 4633 4659 4567 98.03 98.58 98.30 vp 4768 4743 4562 96.18 95.68 95.93 sbar 503 495 460 92.93 91.45 92.18 adjp 384 371 316 85.18 82.29 83.71 advp 822 820 715 87.20 86.98 87.09 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.55 80.48 81.01 Avg2. 23486 23469 22547 96.07 96.00 96.04 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 148 Log-likelihood = -29342.886797 Norm (log-likelihood gradient vector) = 736.100530 Norm (lambda vector) = 347.203793 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4578 98.24 98.81 98.53 i-np 13660 13710 13407 97.79 98.15 97.97 e-np 12220 12220 12029 98.44 98.44 98.44 o 6349 6343 6176 97.37 97.28 97.32 e-vp 4768 4741 4668 98.46 97.90 98.18 i-vp 2602 2638 2522 95.60 96.93 96.26 e-adjp 384 369 333 90.24 86.72 88.45 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 819 720 87.91 87.59 87.75 i-advp 100 89 72 80.90 72.00 76.19 e-sbar 503 494 465 94.13 92.45 93.28 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.58 74.99 75.78 Avg2. 46451 46451 45295 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12220 11771 96.33 96.33 96.33 pp 4633 4660 4567 98.00 98.58 98.29 vp 4768 4741 4562 96.22 95.68 95.95 sbar 503 494 460 93.12 91.45 92.28 adjp 384 369 316 85.64 82.29 83.93 advp 822 819 713 87.06 86.74 86.90 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.67 80.45 81.06 Avg2. 23486 23458 22533 96.06 95.94 96.00 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 149 Log-likelihood = -28907.470243 Norm (log-likelihood gradient vector) = 878.140967 Norm (lambda vector) = 349.755561 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4579 98.18 98.83 98.50 i-np 13660 13711 13410 97.80 98.17 97.99 e-np 12220 12216 12029 98.47 98.44 98.45 o 6349 6343 6178 97.40 97.31 97.35 e-vp 4768 4744 4669 98.42 97.92 98.17 i-vp 2602 2635 2523 95.75 96.96 96.35 e-adjp 384 372 334 89.78 86.98 88.36 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 822 722 87.83 87.83 87.83 i-advp 100 90 72 80.00 72.00 75.79 e-sbar 503 489 461 94.27 91.65 92.94 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 14 11 78.57 91.67 84.62 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.20 74.56 75.86 Avg2. 46451 46451 45301 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12216 11772 96.37 96.33 96.35 pp 4633 4664 4568 97.94 98.60 98.27 vp 4768 4744 4567 96.27 95.78 96.03 sbar 503 489 457 93.46 90.85 92.14 adjp 384 372 317 85.22 82.55 83.86 advp 822 822 715 86.98 86.98 86.98 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.26 79.83 81.02 Avg2. 23486 23460 22539 96.07 95.97 96.02 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 150 Log-likelihood = -28206.039278 Norm (log-likelihood gradient vector) = 911.046949 Norm (lambda vector) = 354.521877 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4658 4575 98.22 98.75 98.48 i-np 13660 13707 13413 97.86 98.19 98.02 e-np 12220 12220 12029 98.44 98.44 98.44 o 6349 6353 6182 97.31 97.37 97.34 e-vp 4768 4741 4667 98.44 97.88 98.16 i-vp 2602 2637 2524 95.71 97.00 96.35 e-adjp 384 370 333 90.00 86.72 88.33 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 821 723 88.06 87.96 88.01 i-advp 100 89 72 80.90 72.00 76.19 e-sbar 503 492 463 94.11 92.05 93.07 i-adjp 152 130 113 86.92 74.34 80.14 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.02 74.51 75.74 Avg2. 46451 46451 45303 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12220 11776 96.37 96.37 96.37 pp 4633 4658 4564 97.98 98.51 98.25 vp 4768 4741 4563 96.25 95.70 95.97 sbar 503 492 458 93.09 91.05 92.06 adjp 384 370 316 85.41 82.29 83.82 advp 822 821 716 87.21 87.10 87.16 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.26 79.82 81.02 Avg2. 23486 23455 22536 96.08 95.96 96.02 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 151 Log-likelihood = -27727.475223 Norm (log-likelihood gradient vector) = 2815.411017 Norm (lambda vector) = 360.476123 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4658 4576 98.24 98.77 98.50 i-np 13660 13706 13413 97.86 98.19 98.03 e-np 12220 12220 12030 98.45 98.45 98.45 o 6349 6347 6180 97.37 97.34 97.35 e-vp 4768 4745 4668 98.38 97.90 98.14 i-vp 2602 2637 2523 95.68 96.96 96.32 e-adjp 384 370 332 89.73 86.46 88.06 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 821 721 87.82 87.71 87.77 i-advp 100 90 72 80.00 72.00 75.79 e-sbar 503 492 464 94.31 92.25 93.27 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.93 74.52 75.71 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12220 11776 96.37 96.37 96.37 pp 4633 4658 4565 98.00 98.53 98.27 vp 4768 4745 4564 96.19 95.72 95.95 sbar 503 492 459 93.29 91.25 92.26 adjp 384 370 315 85.14 82.03 83.55 advp 822 821 714 86.97 86.86 86.91 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.23 79.79 80.99 Avg2. 23486 23459 22536 96.07 95.96 96.01 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 152 Log-likelihood = -27762.336558 Norm (log-likelihood gradient vector) = 1546.298523 Norm (lambda vector) = 357.463303 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4576 98.16 98.77 98.46 i-np 13660 13700 13410 97.88 98.17 98.03 e-np 12220 12224 12033 98.44 98.47 98.45 o 6349 6353 6184 97.34 97.40 97.37 e-vp 4768 4740 4666 98.44 97.86 98.15 i-vp 2602 2638 2524 95.68 97.00 96.34 e-adjp 384 372 333 89.52 86.72 88.10 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 820 722 88.05 87.83 87.94 i-advp 100 90 72 80.00 72.00 75.79 e-sbar 503 488 460 94.26 91.45 92.84 i-adjp 152 132 113 85.61 74.34 79.58 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 14 11 78.57 91.67 84.62 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.15 74.47 75.79 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12224 11780 96.37 96.40 96.38 pp 4633 4662 4565 97.92 98.53 98.22 vp 4768 4740 4561 96.22 95.66 95.94 sbar 503 488 456 93.44 90.66 92.03 adjp 384 372 316 84.95 82.29 83.60 advp 822 820 715 87.20 86.98 87.09 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.24 79.77 80.99 Avg2. 23486 23459 22536 96.07 95.96 96.01 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 153 Log-likelihood = -27339.290834 Norm (log-likelihood gradient vector) = 947.695443 Norm (lambda vector) = 359.647758 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4658 4576 98.24 98.77 98.50 i-np 13660 13708 13409 97.82 98.16 97.99 e-np 12220 12218 12030 98.46 98.45 98.45 o 6349 6354 6183 97.31 97.39 97.35 e-vp 4768 4736 4663 98.46 97.80 98.13 i-vp 2602 2638 2524 95.68 97.00 96.34 e-adjp 384 376 335 89.10 87.24 88.16 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 818 721 88.14 87.71 87.93 i-advp 100 86 71 82.56 71.00 76.34 e-sbar 503 492 463 94.11 92.05 93.07 i-adjp 152 133 114 85.71 75.00 80.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.41 74.41 75.39 Avg2. 46451 46451 45297 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12218 11777 96.39 96.37 96.38 pp 4633 4658 4565 98.00 98.53 98.27 vp 4768 4736 4558 96.24 95.60 95.92 sbar 503 492 458 93.09 91.05 92.06 adjp 384 376 318 84.57 82.81 83.68 advp 822 818 715 87.41 86.98 87.20 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.50 79.85 80.67 Avg2. 23486 23452 22534 96.09 95.95 96.02 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 154 Log-likelihood = -26766.896420 Norm (log-likelihood gradient vector) = 718.449761 Norm (lambda vector) = 362.800292 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4579 98.20 98.83 98.52 i-np 13660 13690 13406 97.93 98.14 98.03 e-np 12220 12229 12039 98.45 98.52 98.48 o 6349 6363 6190 97.28 97.50 97.39 e-vp 4768 4735 4663 98.48 97.80 98.14 i-vp 2602 2638 2524 95.68 97.00 96.34 e-adjp 384 376 335 89.10 87.24 88.16 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 816 721 88.36 87.71 88.03 i-advp 100 85 71 83.53 71.00 76.76 e-sbar 503 489 461 94.27 91.65 92.94 i-adjp 152 133 114 85.71 75.00 80.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.48 74.40 75.42 Avg2. 46451 46451 45311 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12229 11790 96.41 96.48 96.45 pp 4633 4663 4568 97.96 98.60 98.28 vp 4768 4735 4558 96.26 95.60 95.93 sbar 503 489 456 93.25 90.66 91.94 adjp 384 376 318 84.57 82.81 83.68 advp 822 816 715 87.62 86.98 87.30 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.54 79.83 80.67 Avg2. 23486 23462 22548 96.10 96.01 96.06 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 155 Log-likelihood = -26376.187709 Norm (log-likelihood gradient vector) = 1021.512903 Norm (lambda vector) = 365.374781 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4577 98.24 98.79 98.51 i-np 13660 13804 13446 97.41 98.43 97.92 e-np 12220 12186 12010 98.56 98.28 98.42 o 6349 6302 6153 97.64 96.91 97.27 e-vp 4768 4732 4663 98.54 97.80 98.17 i-vp 2602 2640 2527 95.72 97.12 96.41 e-adjp 384 373 333 89.28 86.72 87.98 i-pp 52 44 34 77.27 65.38 70.83 e-advp 822 812 720 88.67 87.59 88.13 i-advp 100 84 71 84.52 71.00 77.17 e-sbar 503 492 463 94.11 92.05 93.07 i-adjp 152 130 111 85.38 73.03 78.72 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.94 74.26 75.09 Avg2. 46451 46451 45282 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12186 11751 96.43 96.16 96.30 pp 4633 4659 4566 98.00 98.55 98.28 vp 4768 4732 4557 96.30 95.57 95.94 sbar 503 492 458 93.09 91.05 92.06 adjp 384 373 316 84.72 82.29 83.49 advp 822 812 714 87.93 86.86 87.39 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.97 79.77 80.36 Avg2. 23486 23409 22505 96.14 95.82 95.98 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 156 Log-likelihood = -26012.021769 Norm (log-likelihood gradient vector) = 2401.301193 Norm (lambda vector) = 369.947650 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4579 98.20 98.83 98.52 i-np 13660 13719 13415 97.78 98.21 97.99 e-np 12220 12223 12034 98.45 98.48 98.47 o 6349 6340 6181 97.49 97.35 97.42 e-vp 4768 4734 4664 98.52 97.82 98.17 i-vp 2602 2643 2529 95.69 97.19 96.43 e-adjp 384 377 334 88.59 86.98 87.78 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 815 722 88.59 87.83 88.21 i-advp 100 84 71 84.52 71.00 77.17 e-sbar 503 489 462 94.48 91.85 93.15 i-adjp 152 130 111 85.38 73.03 78.72 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.51 74.31 75.39 Avg2. 46451 46451 45310 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12223 11783 96.40 96.42 96.41 pp 4633 4663 4568 97.96 98.60 98.28 vp 4768 4734 4559 96.30 95.62 95.96 sbar 503 489 457 93.46 90.85 92.14 adjp 384 377 317 84.08 82.55 83.31 advp 822 815 716 87.85 87.10 87.48 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.53 79.83 80.67 Avg2. 23486 23455 22543 96.11 95.98 96.05 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 157 Log-likelihood = -25714.468828 Norm (log-likelihood gradient vector) = 719.067821 Norm (lambda vector) = 370.045388 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4661 4579 98.24 98.83 98.54 i-np 13660 13690 13401 97.89 98.10 98.00 e-np 12220 12238 12042 98.40 98.54 98.47 o 6349 6353 6189 97.42 97.48 97.45 e-vp 4768 4739 4667 98.48 97.88 98.18 i-vp 2602 2646 2530 95.62 97.23 96.42 e-adjp 384 373 334 89.54 86.98 88.24 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 815 722 88.59 87.83 88.21 i-advp 100 84 71 84.52 71.00 77.17 e-sbar 503 489 463 94.68 92.05 93.35 i-adjp 152 130 112 86.15 73.68 79.43 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 15 11 73.33 91.67 81.48 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.70 74.36 75.51 Avg2. 46451 46451 45318 97.56 97.56 97.56 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12238 11793 96.36 96.51 96.43 pp 4633 4661 4568 98.00 98.60 98.30 vp 4768 4739 4561 96.24 95.66 95.95 sbar 503 489 458 93.66 91.05 92.34 adjp 384 373 317 84.99 82.55 83.75 advp 822 815 716 87.85 87.10 87.48 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.64 79.86 80.74 Avg2. 23486 23469 22556 96.11 96.04 96.07 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 158 Log-likelihood = -25582.036107 Norm (log-likelihood gradient vector) = 681.103714 Norm (lambda vector) = 370.605319 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4577 98.28 98.79 98.54 i-np 13660 13680 13398 97.94 98.08 98.01 e-np 12220 12240 12046 98.42 98.58 98.50 o 6349 6356 6190 97.39 97.50 97.44 e-vp 4768 4740 4667 98.46 97.88 98.17 i-vp 2602 2646 2530 95.62 97.23 96.42 e-adjp 384 372 334 89.78 86.98 88.36 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 817 722 88.37 87.83 88.10 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 494 466 94.33 92.64 93.48 i-adjp 152 131 113 86.26 74.34 79.86 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.51 74.05 75.26 Avg2. 46451 46451 45322 97.57 97.57 97.57 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12240 11798 96.39 96.55 96.47 pp 4633 4657 4565 98.02 98.53 98.28 vp 4768 4740 4561 96.22 95.66 95.94 sbar 503 494 460 93.12 91.45 92.28 adjp 384 372 317 85.22 82.55 83.86 advp 822 817 716 87.64 87.10 87.37 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.59 79.90 80.74 Avg2. 23486 23474 22560 96.11 96.06 96.08 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 159 Log-likelihood = -25187.576115 Norm (log-likelihood gradient vector) = 817.370849 Norm (lambda vector) = 373.637238 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4672 4583 98.10 98.92 98.51 i-np 13660 13675 13381 97.85 97.96 97.90 e-np 12220 12245 12039 98.32 98.52 98.42 o 6349 6355 6182 97.28 97.37 97.32 e-vp 4768 4736 4663 98.46 97.80 98.13 i-vp 2602 2643 2527 95.61 97.12 96.36 e-adjp 384 375 334 89.07 86.98 88.01 i-pp 52 43 35 81.40 67.31 73.68 e-advp 822 819 722 88.16 87.83 88.00 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 481 458 95.22 91.05 93.09 i-adjp 152 132 113 85.61 74.34 79.58 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.52 74.01 75.24 Avg2. 46451 46451 45281 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12245 11780 96.20 96.40 96.30 pp 4633 4672 4571 97.84 98.66 98.25 vp 4768 4736 4557 96.22 95.57 95.90 sbar 503 481 452 93.97 89.86 91.87 adjp 384 375 317 84.53 82.55 83.53 advp 822 819 716 87.42 87.10 87.26 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.54 79.65 80.59 Avg2. 23486 23481 22535 95.97 95.95 95.96 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 160 Log-likelihood = -24716.839316 Norm (log-likelihood gradient vector) = 1736.489870 Norm (lambda vector) = 379.030728 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4579 98.20 98.83 98.52 i-np 13660 13680 13396 97.92 98.07 98.00 e-np 12220 12240 12045 98.41 98.57 98.49 o 6349 6352 6186 97.39 97.43 97.41 e-vp 4768 4737 4663 98.44 97.80 98.12 i-vp 2602 2642 2526 95.61 97.08 96.34 e-adjp 384 376 334 88.83 86.98 87.89 i-pp 52 42 34 80.95 65.38 72.34 e-advp 822 822 723 87.96 87.96 87.96 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 488 463 94.88 92.05 93.44 i-adjp 152 132 113 85.61 74.34 79.58 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.97 73.98 74.96 Avg2. 46451 46451 45306 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12240 11793 96.35 96.51 96.43 pp 4633 4663 4567 97.94 98.58 98.26 vp 4768 4737 4556 96.18 95.55 95.87 sbar 503 488 457 93.65 90.85 92.23 adjp 384 376 317 84.31 82.55 83.42 advp 822 822 717 87.23 87.23 87.23 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.88 79.76 80.32 Avg2. 23486 23480 22549 96.03 96.01 96.02 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 161 Log-likelihood = -24243.845827 Norm (log-likelihood gradient vector) = 763.241074 Norm (lambda vector) = 381.627158 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4575 98.24 98.75 98.49 i-np 13660 13691 13404 97.90 98.13 98.01 e-np 12220 12236 12044 98.43 98.56 98.50 o 6349 6345 6182 97.43 97.37 97.40 e-vp 4768 4736 4663 98.46 97.80 98.13 i-vp 2602 2643 2527 95.61 97.12 96.36 e-adjp 384 378 336 88.89 87.50 88.19 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 818 723 88.39 87.96 88.17 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 493 465 94.32 92.45 93.37 i-adjp 152 136 116 85.29 76.32 80.56 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.97 74.12 75.03 Avg2. 46451 46451 45313 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12236 11793 96.38 96.51 96.44 pp 4633 4657 4563 97.98 98.49 98.23 vp 4768 4736 4557 96.22 95.57 95.90 sbar 503 493 459 93.10 91.25 92.17 adjp 384 378 318 84.13 82.81 83.46 advp 822 818 717 87.65 87.23 87.44 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.87 79.82 80.34 Avg2. 23486 23472 22549 96.07 96.01 96.04 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 162 Log-likelihood = -23907.068630 Norm (log-likelihood gradient vector) = 589.589802 Norm (lambda vector) = 383.866523 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4659 4576 98.22 98.77 98.49 i-np 13660 13702 13406 97.84 98.14 97.99 e-np 12220 12228 12040 98.46 98.53 98.49 o 6349 6347 6179 97.35 97.32 97.34 e-vp 4768 4734 4661 98.46 97.76 98.11 i-vp 2602 2640 2524 95.61 97.00 96.30 e-adjp 384 378 336 88.89 87.50 88.19 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 817 722 88.37 87.83 88.10 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 492 464 94.31 92.25 93.27 i-adjp 152 136 116 85.29 76.32 80.56 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.96 74.09 75.01 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11788 96.40 96.46 96.43 pp 4633 4659 4564 97.96 98.51 98.24 vp 4768 4734 4552 96.16 95.47 95.81 sbar 503 492 458 93.09 91.05 92.06 adjp 384 378 319 84.39 83.07 83.73 advp 822 817 716 87.64 87.10 87.37 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.88 79.80 80.34 Avg2. 23486 23462 22539 96.07 95.97 96.02 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 163 Log-likelihood = -23535.053620 Norm (log-likelihood gradient vector) = 692.579959 Norm (lambda vector) = 386.860219 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4653 4571 98.24 98.66 98.45 i-np 13660 13760 13428 97.59 98.30 97.94 e-np 12220 12205 12022 98.50 98.38 98.44 o 6349 6319 6165 97.56 97.10 97.33 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2642 2525 95.57 97.04 96.30 e-adjp 384 376 335 89.10 87.24 88.16 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 816 719 88.11 87.47 87.79 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 496 464 93.55 92.25 92.89 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.96 74.01 74.97 Avg2. 46451 46451 45283 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12205 11759 96.35 96.23 96.29 pp 4633 4653 4559 97.98 98.40 98.19 vp 4768 4733 4552 96.18 95.47 95.82 sbar 503 496 458 92.34 91.05 91.69 adjp 384 376 318 84.57 82.81 83.68 advp 822 816 713 87.38 86.74 87.06 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.80 79.71 80.25 Avg2. 23486 23433 22501 96.02 95.81 95.91 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 164 Log-likelihood = -23037.943552 Norm (log-likelihood gradient vector) = 2179.520632 Norm (lambda vector) = 392.424508 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4660 4575 98.18 98.75 98.46 i-np 13660 13700 13407 97.86 98.15 98.00 e-np 12220 12229 12038 98.44 98.51 98.47 o 6349 6347 6181 97.38 97.35 97.37 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2643 2526 95.57 97.08 96.32 e-adjp 384 376 334 88.83 86.98 87.89 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 816 721 88.36 87.71 88.03 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 494 463 93.72 92.05 92.88 i-adjp 152 134 116 86.57 76.32 81.12 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.94 74.05 74.98 Avg2. 46451 46451 45300 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12229 11787 96.39 96.46 96.42 pp 4633 4660 4563 97.92 98.49 98.20 vp 4768 4733 4553 96.20 95.49 95.84 sbar 503 494 457 92.51 90.85 91.68 adjp 384 376 318 84.57 82.81 83.68 advp 822 816 715 87.62 86.98 87.30 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.84 79.75 80.29 Avg2. 23486 23462 22535 96.05 95.95 96.00 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 165 Log-likelihood = -22720.843148 Norm (log-likelihood gradient vector) = 715.724038 Norm (lambda vector) = 394.305955 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4577 98.18 98.79 98.48 i-np 13660 13693 13407 97.91 98.15 98.03 e-np 12220 12233 12038 98.41 98.51 98.46 o 6349 6349 6180 97.34 97.34 97.34 e-vp 4768 4734 4662 98.48 97.78 98.13 i-vp 2602 2642 2526 95.61 97.08 96.34 e-adjp 384 377 335 88.86 87.24 88.04 i-pp 52 43 34 79.07 65.38 71.58 e-advp 822 815 720 88.34 87.59 87.97 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 493 463 93.91 92.05 92.97 i-adjp 152 134 116 86.57 76.32 81.12 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 23 20 86.96 83.33 85.11 e-conjp 16 14 11 78.57 68.75 73.33 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.95 74.06 74.99 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12233 11787 96.35 96.46 96.41 pp 4633 4662 4565 97.92 98.53 98.22 vp 4768 4734 4555 96.22 95.53 95.87 sbar 503 493 457 92.70 90.85 91.77 adjp 384 377 318 84.35 82.81 83.57 advp 822 815 714 87.61 86.86 87.23 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 11 78.57 68.75 73.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.83 79.74 80.28 Avg2. 23486 23468 22538 96.04 95.96 96.00 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 166 Log-likelihood = -22638.856131 Norm (log-likelihood gradient vector) = 629.421436 Norm (lambda vector) = 394.487854 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4580 98.07 98.86 98.46 i-np 13660 13697 13410 97.90 98.17 98.04 e-np 12220 12235 12040 98.41 98.53 98.47 o 6349 6343 6179 97.41 97.32 97.37 e-vp 4768 4732 4659 98.46 97.71 98.08 i-vp 2602 2644 2526 95.54 97.08 96.30 e-adjp 384 378 334 88.36 86.98 87.66 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 814 719 88.33 87.47 87.90 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 486 458 94.24 91.05 92.62 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.32 74.03 75.16 Avg2. 46451 46451 45298 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12235 11790 96.36 96.48 96.42 pp 4633 4670 4567 97.79 98.58 98.18 vp 4768 4732 4552 96.20 95.47 95.83 sbar 503 486 452 93.00 89.86 91.41 adjp 384 378 317 83.86 82.55 83.20 advp 822 814 713 87.59 86.74 87.16 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.33 79.60 80.46 Avg2. 23486 23469 22533 96.01 95.94 95.98 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 167 Log-likelihood = -22439.418803 Norm (log-likelihood gradient vector) = 759.570376 Norm (lambda vector) = 395.940372 Log-likelihood and gradient computational time: 322 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4577 98.07 98.79 98.43 i-np 13660 13702 13403 97.82 98.12 97.97 e-np 12220 12228 12038 98.45 98.51 98.48 o 6349 6341 6179 97.45 97.32 97.38 e-vp 4768 4735 4660 98.42 97.73 98.07 i-vp 2602 2644 2526 95.54 97.08 96.30 e-adjp 384 378 334 88.36 86.98 87.66 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 816 720 88.24 87.59 87.91 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 488 460 94.26 91.45 92.84 i-adjp 152 132 114 86.36 75.00 80.28 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.31 74.05 75.16 Avg2. 46451 46451 45290 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11783 96.36 96.42 96.39 pp 4633 4667 4564 97.79 98.51 98.15 vp 4768 4735 4553 96.16 95.49 95.82 sbar 503 488 454 93.03 90.26 91.62 adjp 384 378 317 83.86 82.55 83.20 advp 822 816 714 87.50 86.86 87.18 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.32 79.65 80.48 Avg2. 23486 23466 22527 96.00 95.92 95.96 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 357 seconds Iteration: 168 Log-likelihood = -22180.742388 Norm (log-likelihood gradient vector) = 773.198488 Norm (lambda vector) = 398.160013 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4575 98.09 98.75 98.42 i-np 13660 13709 13397 97.72 98.07 97.90 e-np 12220 12224 12032 98.43 98.46 98.45 o 6349 6338 6173 97.40 97.23 97.31 e-vp 4768 4733 4659 98.44 97.71 98.07 i-vp 2602 2644 2527 95.57 97.12 96.34 e-adjp 384 380 336 88.42 87.50 87.96 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 812 719 88.55 87.47 88.00 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 490 460 93.88 91.45 92.65 i-adjp 152 135 117 86.67 76.97 81.53 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.28 74.19 75.22 Avg2. 46451 46451 45275 97.47 97.47 97.47 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12224 11765 96.25 96.28 96.26 pp 4633 4664 4562 97.81 98.47 98.14 vp 4768 4733 4554 96.22 95.51 95.86 sbar 503 490 454 92.65 90.26 91.44 adjp 384 380 320 84.21 83.33 83.77 advp 822 812 712 87.68 86.62 87.15 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.34 79.76 80.54 Avg2. 23486 23458 22510 95.96 95.84 95.90 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 169 Log-likelihood = -21797.953283 Norm (log-likelihood gradient vector) = 1325.556851 Norm (lambda vector) = 404.056713 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4573 98.05 98.70 98.38 i-np 13660 13724 13419 97.78 98.24 98.01 e-np 12220 12221 12037 98.49 98.50 98.50 o 6349 6334 6177 97.52 97.29 97.41 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2648 2531 95.58 97.27 96.42 e-adjp 384 377 335 88.86 87.24 88.04 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 808 717 88.74 87.23 87.98 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 489 457 93.46 90.85 92.14 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.27 73.71 74.97 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12221 11783 96.42 96.42 96.42 pp 4633 4664 4560 97.77 98.42 98.10 vp 4768 4733 4559 96.32 95.62 95.97 sbar 503 489 450 92.02 89.46 90.73 adjp 384 377 319 84.62 83.07 83.84 advp 822 808 710 87.87 86.37 87.12 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.36 79.65 80.50 Avg2. 23486 23447 22524 96.06 95.90 95.98 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 170 Log-likelihood = -21461.371463 Norm (log-likelihood gradient vector) = 767.891076 Norm (lambda vector) = 408.747878 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4573 98.09 98.70 98.40 i-np 13660 13710 13413 97.83 98.19 98.01 e-np 12220 12227 12041 98.48 98.54 98.51 o 6349 6341 6181 97.48 97.35 97.42 e-vp 4768 4732 4661 98.50 97.76 98.13 i-vp 2602 2649 2531 95.55 97.27 96.40 e-adjp 384 377 335 88.86 87.24 88.04 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 807 716 88.72 87.10 87.91 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 493 460 93.31 91.45 92.37 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.26 73.74 74.98 Avg2. 46451 46451 45306 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12227 11787 96.40 96.46 96.43 pp 4633 4662 4560 97.81 98.42 98.12 vp 4768 4732 4558 96.32 95.60 95.96 sbar 503 493 453 91.89 90.06 90.96 adjp 384 377 319 84.62 83.07 83.84 advp 822 807 709 87.86 86.25 87.05 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.34 79.70 80.52 Avg2. 23486 23453 22529 96.06 95.93 95.99 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 171 Log-likelihood = -21337.110925 Norm (log-likelihood gradient vector) = 492.199058 Norm (lambda vector) = 408.595607 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4662 4574 98.11 98.73 98.42 i-np 13660 13713 13416 97.83 98.21 98.02 e-np 12220 12225 12039 98.48 98.52 98.50 o 6349 6342 6183 97.49 97.39 97.44 e-vp 4768 4732 4659 98.46 97.71 98.08 i-vp 2602 2646 2529 95.58 97.19 96.38 e-adjp 384 376 334 88.83 86.98 87.89 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 809 717 88.63 87.23 87.92 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 494 461 93.32 91.65 92.48 i-adjp 152 134 117 87.31 76.97 81.82 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.39 73.77 75.05 Avg2. 46451 46451 45308 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12225 11786 96.41 96.45 96.43 pp 4633 4662 4562 97.85 98.47 98.16 vp 4768 4732 4555 96.26 95.53 95.89 sbar 503 494 454 91.90 90.26 91.07 adjp 384 376 318 84.57 82.81 83.68 advp 822 809 710 87.76 86.37 87.06 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.40 79.71 80.55 Avg2. 23486 23452 22528 96.06 95.92 95.99 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 172 Log-likelihood = -21177.611523 Norm (log-likelihood gradient vector) = 620.937535 Norm (lambda vector) = 409.697195 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4576 98.09 98.77 98.43 i-np 13660 13610 13352 98.10 97.75 97.92 e-np 12220 12271 12055 98.24 98.65 98.44 o 6349 6390 6199 97.01 97.64 97.32 e-vp 4768 4735 4658 98.37 97.69 98.03 i-vp 2602 2652 2527 95.29 97.12 96.19 e-adjp 384 375 332 88.53 86.46 87.48 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 806 715 88.71 86.98 87.84 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 492 461 93.70 91.65 92.66 i-adjp 152 135 116 85.93 76.32 80.84 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.20 73.69 74.92 Avg2. 46451 46451 45270 97.46 97.46 97.46 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12271 11793 96.10 96.51 96.30 pp 4633 4665 4564 97.83 98.51 98.17 vp 4768 4735 4550 96.09 95.43 95.76 sbar 503 492 454 92.28 90.26 91.26 adjp 384 375 316 84.27 82.29 83.27 advp 822 806 708 87.84 86.13 86.98 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.30 79.63 80.45 Avg2. 23486 23499 22528 95.87 95.92 95.89 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 173 Log-likelihood = -21235.817566 Norm (log-likelihood gradient vector) = 2708.640428 Norm (lambda vector) = 413.989311 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4575 98.11 98.75 98.43 i-np 13660 13668 13384 97.92 97.98 97.95 e-np 12220 12246 12048 98.38 98.59 98.49 o 6349 6363 6190 97.28 97.50 97.39 e-vp 4768 4735 4659 98.39 97.71 98.05 i-vp 2602 2644 2526 95.54 97.08 96.30 e-adjp 384 376 333 88.56 86.72 87.63 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 810 717 88.52 87.23 87.87 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 492 461 93.70 91.65 92.66 i-adjp 152 135 117 86.67 76.97 81.53 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.29 73.75 75.00 Avg2. 46451 46451 45289 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12246 11789 96.27 96.47 96.37 pp 4633 4663 4563 97.86 98.49 98.17 vp 4768 4735 4553 96.16 95.49 95.82 sbar 503 492 454 92.28 90.26 91.26 adjp 384 376 317 84.31 82.55 83.42 advp 822 810 710 87.65 86.37 87.01 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.38 79.68 80.52 Avg2. 23486 23476 22529 95.97 95.93 95.95 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 174 Log-likelihood = -21114.015460 Norm (log-likelihood gradient vector) = 1097.158433 Norm (lambda vector) = 411.456153 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4576 98.11 98.77 98.44 i-np 13660 13681 13392 97.89 98.04 97.96 e-np 12220 12241 12045 98.40 98.57 98.48 o 6349 6357 6188 97.34 97.46 97.40 e-vp 4768 4733 4658 98.42 97.69 98.05 i-vp 2602 2650 2527 95.36 97.12 96.23 e-adjp 384 373 332 89.01 86.46 87.71 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 808 716 88.61 87.10 87.85 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 492 461 93.70 91.65 92.66 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.34 73.70 74.99 Avg2. 46451 46451 45290 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12241 11787 96.29 96.46 96.37 pp 4633 4664 4564 97.86 98.51 98.18 vp 4768 4733 4550 96.13 95.43 95.78 sbar 503 492 454 92.28 90.26 91.26 adjp 384 373 316 84.72 82.29 83.49 advp 822 808 709 87.75 86.25 86.99 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.43 79.64 80.52 Avg2. 23486 23465 22523 95.99 95.90 95.94 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 175 Log-likelihood = -20972.772968 Norm (log-likelihood gradient vector) = 874.806070 Norm (lambda vector) = 413.639505 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4577 98.07 98.79 98.43 i-np 13660 13706 13410 97.84 98.17 98.00 e-np 12220 12231 12040 98.44 98.53 98.48 o 6349 6335 6180 97.55 97.34 97.45 e-vp 4768 4731 4657 98.44 97.67 98.05 i-vp 2602 2651 2528 95.36 97.16 96.25 e-adjp 384 375 333 88.80 86.72 87.75 i-pp 52 46 35 76.09 67.31 71.43 e-advp 822 808 717 88.74 87.23 87.98 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 492 461 93.70 91.65 92.66 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.24 74.10 75.15 Avg2. 46451 46451 45298 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12231 11784 96.35 96.43 96.39 pp 4633 4667 4564 97.79 98.51 98.15 vp 4768 4731 4549 96.15 95.41 95.78 sbar 503 492 455 92.48 90.46 91.46 adjp 384 375 316 84.27 82.29 83.27 advp 822 808 710 87.87 86.37 87.12 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.35 79.66 80.50 Avg2. 23486 23459 22521 96.00 95.89 95.95 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 176 Log-likelihood = -20636.663800 Norm (log-likelihood gradient vector) = 604.753556 Norm (lambda vector) = 419.445173 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4672 4579 98.01 98.83 98.42 i-np 13660 13716 13408 97.75 98.16 97.95 e-np 12220 12229 12034 98.41 98.48 98.44 o 6349 6331 6176 97.55 97.28 97.41 e-vp 4768 4734 4659 98.42 97.71 98.06 i-vp 2602 2647 2527 95.47 97.12 96.29 e-adjp 384 375 333 88.80 86.72 87.75 i-pp 52 46 35 76.09 67.31 71.43 e-advp 822 807 715 88.60 86.98 87.78 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 486 457 94.03 90.85 92.42 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.24 74.01 75.11 Avg2. 46451 46451 45282 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12229 11773 96.27 96.34 96.31 pp 4633 4672 4566 97.73 98.55 98.14 vp 4768 4734 4552 96.16 95.47 95.81 sbar 503 486 451 92.80 89.66 91.20 adjp 384 375 316 84.27 82.29 83.27 advp 822 807 708 87.73 86.13 86.92 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.35 79.48 80.40 Avg2. 23486 23457 22508 95.95 95.84 95.90 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 177 Log-likelihood = -20403.283254 Norm (log-likelihood gradient vector) = 631.485766 Norm (lambda vector) = 423.754629 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4669 4578 98.05 98.81 98.43 i-np 13660 13716 13405 97.73 98.13 97.93 e-np 12220 12228 12032 98.40 98.46 98.43 o 6349 6332 6175 97.52 97.26 97.39 e-vp 4768 4734 4658 98.39 97.69 98.04 i-vp 2602 2646 2525 95.43 97.04 96.23 e-adjp 384 377 334 88.59 86.98 87.78 i-pp 52 46 35 76.09 67.31 71.43 e-advp 822 808 715 88.49 86.98 87.73 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 488 458 93.85 91.05 92.43 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 20 90.91 83.33 86.96 e-conjp 16 13 11 84.62 68.75 75.86 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.10 73.60 74.83 Avg2. 46451 46451 45273 97.46 97.46 97.46 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11768 96.24 96.30 96.27 pp 4633 4669 4565 97.77 98.53 98.15 vp 4768 4734 4549 96.09 95.41 95.75 sbar 503 488 451 92.42 89.66 91.02 adjp 384 377 317 84.08 82.55 83.31 advp 822 808 708 87.62 86.13 86.87 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 11 84.62 68.75 75.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.27 79.50 80.37 Avg2. 23486 23458 22500 95.92 95.80 95.86 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 178 Log-likelihood = -20204.181710 Norm (log-likelihood gradient vector) = 534.240866 Norm (lambda vector) = 427.600771 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4577 98.11 98.79 98.45 i-np 13660 13714 13409 97.78 98.16 97.97 e-np 12220 12230 12035 98.41 98.49 98.45 o 6349 6335 6178 97.52 97.31 97.41 e-vp 4768 4735 4658 98.37 97.69 98.03 i-vp 2602 2645 2524 95.43 97.00 96.21 e-adjp 384 376 334 88.83 86.98 87.89 i-pp 52 47 36 76.60 69.23 72.73 e-advp 822 809 716 88.50 87.10 87.80 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 491 461 93.89 91.65 92.76 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.76 73.74 75.22 Avg2. 46451 46451 45286 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12230 11775 96.28 96.36 96.32 pp 4633 4665 4565 97.86 98.53 98.19 vp 4768 4735 4548 96.05 95.39 95.72 sbar 503 491 454 92.46 90.26 91.35 adjp 384 376 317 84.31 82.55 83.42 advp 822 809 709 87.64 86.25 86.94 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.02 79.57 80.77 Avg2. 23486 23459 22510 95.95 95.84 95.90 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 179 Log-likelihood = -19974.598018 Norm (log-likelihood gradient vector) = 529.865122 Norm (lambda vector) = 430.442900 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4668 4579 98.09 98.83 98.46 i-np 13660 13710 13403 97.76 98.12 97.94 e-np 12220 12228 12033 98.41 98.47 98.44 o 6349 6334 6175 97.49 97.26 97.37 e-vp 4768 4733 4658 98.42 97.69 98.05 i-vp 2602 2646 2526 95.46 97.08 96.27 e-adjp 384 380 336 88.42 87.50 87.96 i-pp 52 47 36 76.60 69.23 72.73 e-advp 822 810 717 88.52 87.23 87.87 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 491 460 93.69 91.45 92.56 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.79 73.79 75.26 Avg2. 46451 46451 45282 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12228 11773 96.28 96.34 96.31 pp 4633 4668 4567 97.84 98.58 98.20 vp 4768 4733 4550 96.13 95.43 95.78 sbar 503 491 453 92.26 90.06 91.15 adjp 384 380 319 83.95 83.07 83.51 advp 822 810 710 87.65 86.37 87.01 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.97 79.62 80.78 Avg2. 23486 23463 22514 95.96 95.86 95.91 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 180 Log-likelihood = -19805.778402 Norm (log-likelihood gradient vector) = 1072.552431 Norm (lambda vector) = 435.595425 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4663 4577 98.16 98.79 98.47 i-np 13660 13705 13403 97.80 98.12 97.96 e-np 12220 12234 12037 98.39 98.50 98.45 o 6349 6337 6177 97.48 97.29 97.38 e-vp 4768 4730 4657 98.46 97.67 98.06 i-vp 2602 2647 2526 95.43 97.08 96.25 e-adjp 384 380 336 88.42 87.50 87.96 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 808 716 88.61 87.10 87.85 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 495 463 93.54 92.05 92.79 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.92 74.34 75.61 Avg2. 46451 46451 45289 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12234 11776 96.26 96.37 96.31 pp 4633 4663 4565 97.90 98.53 98.21 vp 4768 4730 4548 96.15 95.39 95.77 sbar 503 495 456 92.12 90.66 91.38 adjp 384 380 319 83.95 83.07 83.51 advp 822 808 709 87.75 86.25 86.99 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.03 80.29 81.15 Avg2. 23486 23464 22516 95.96 95.87 95.91 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 181 Log-likelihood = -19527.193768 Norm (log-likelihood gradient vector) = 576.271519 Norm (lambda vector) = 437.080053 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4579 98.16 98.83 98.49 i-np 13660 13701 13400 97.80 98.10 97.95 e-np 12220 12237 12038 98.37 98.51 98.44 o 6349 6341 6177 97.41 97.29 97.35 e-vp 4768 4728 4656 98.48 97.65 98.06 i-vp 2602 2648 2527 95.43 97.12 96.27 e-adjp 384 378 335 88.62 87.24 87.93 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 807 716 88.72 87.10 87.91 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 492 462 93.90 91.85 92.86 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 131 123 93.89 97.62 95.72 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.41 74.32 75.35 Avg2. 46451 46451 45287 97.49 97.49 97.49 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12237 11776 96.23 96.37 96.30 pp 4633 4665 4567 97.90 98.58 98.24 vp 4768 4728 4547 96.17 95.36 95.77 sbar 503 492 455 92.48 90.46 91.46 adjp 384 378 318 84.13 82.81 83.46 advp 822 807 709 87.86 86.25 87.05 prt 126 131 123 93.89 97.62 95.72 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.44 80.24 80.84 Avg2. 23486 23462 22515 95.96 95.87 95.91 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 182 Log-likelihood = -19383.607573 Norm (log-likelihood gradient vector) = 518.321040 Norm (lambda vector) = 437.007149 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4581 98.16 98.88 98.52 i-np 13660 13740 13413 97.62 98.19 97.91 e-np 12220 12214 12024 98.44 98.40 98.42 o 6349 6324 6165 97.49 97.10 97.29 e-vp 4768 4729 4655 98.44 97.63 98.03 i-vp 2602 2648 2525 95.35 97.04 96.19 e-adjp 384 378 336 88.89 87.50 88.19 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 808 716 88.61 87.10 87.85 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 489 461 94.27 91.65 92.94 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 130 123 94.62 97.62 96.09 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.42 74.31 75.35 Avg2. 46451 46451 45273 97.46 97.46 97.46 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12214 11759 96.27 96.23 96.25 pp 4633 4667 4569 97.90 98.62 98.26 vp 4768 4729 4544 96.09 95.30 95.69 sbar 503 489 454 92.84 90.26 91.53 adjp 384 378 319 84.39 83.07 83.73 advp 822 808 709 87.75 86.25 86.99 prt 126 130 123 94.62 97.62 96.09 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.56 80.24 80.89 Avg2. 23486 23439 22497 95.98 95.79 95.88 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 183 Log-likelihood = -19283.404163 Norm (log-likelihood gradient vector) = 1260.080680 Norm (lambda vector) = 436.879341 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4580 98.14 98.86 98.49 i-np 13660 13697 13401 97.84 98.10 97.97 e-np 12220 12233 12038 98.41 98.51 98.46 o 6349 6343 6180 97.43 97.34 97.38 e-vp 4768 4733 4659 98.44 97.71 98.07 i-vp 2602 2647 2526 95.43 97.08 96.25 e-adjp 384 380 337 88.68 87.76 88.22 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 807 715 88.60 86.98 87.78 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 489 461 94.27 91.65 92.94 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 23 21 91.30 87.50 89.36 e-conjp 16 14 12 85.71 75.00 80.00 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.42 74.37 75.38 Avg2. 46451 46451 45295 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12233 11780 96.30 96.40 96.35 pp 4633 4667 4568 97.88 98.60 98.24 vp 4768 4733 4549 96.11 95.41 95.76 sbar 503 489 454 92.84 90.26 91.53 adjp 384 380 320 84.21 83.33 83.77 advp 822 807 708 87.73 86.13 86.92 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 14 12 85.71 75.00 80.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.54 80.35 80.94 Avg2. 23486 23464 22523 95.99 95.90 95.94 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 184 Log-likelihood = -19173.704765 Norm (log-likelihood gradient vector) = 472.658444 Norm (lambda vector) = 438.120592 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4579 98.16 98.83 98.49 i-np 13660 13691 13399 97.87 98.09 97.98 e-np 12220 12238 12040 98.38 98.53 98.45 o 6349 6345 6183 97.45 97.39 97.42 e-vp 4768 4733 4659 98.44 97.71 98.07 i-vp 2602 2647 2526 95.43 97.08 96.25 e-adjp 384 381 337 88.45 87.76 88.10 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 808 715 88.49 86.98 87.73 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 490 461 94.08 91.65 92.85 i-adjp 152 133 116 87.22 76.32 81.40 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.93 74.37 75.63 Avg2. 46451 46451 45297 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12238 11783 96.28 96.42 96.35 pp 4633 4665 4567 97.90 98.58 98.24 vp 4768 4733 4549 96.11 95.41 95.76 sbar 503 490 454 92.65 90.26 91.44 adjp 384 381 320 83.99 83.33 83.66 advp 822 808 708 87.62 86.13 86.87 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.15 80.35 81.24 Avg2. 23486 23469 22525 95.98 95.91 95.94 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 185 Log-likelihood = -19084.509422 Norm (log-likelihood gradient vector) = 525.250148 Norm (lambda vector) = 439.873433 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4578 98.14 98.81 98.47 i-np 13660 13696 13401 97.85 98.10 97.97 e-np 12220 12235 12037 98.38 98.50 98.44 o 6349 6338 6181 97.52 97.35 97.44 e-vp 4768 4737 4662 98.42 97.78 98.10 i-vp 2602 2646 2526 95.46 97.08 96.27 e-adjp 384 382 338 88.48 88.02 88.25 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 811 716 88.29 87.10 87.69 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 489 460 94.07 91.45 92.74 i-adjp 152 132 115 87.12 75.66 80.99 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.92 74.35 75.61 Avg2. 46451 46451 45296 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12235 11779 96.27 96.39 96.33 pp 4633 4665 4566 97.88 98.55 98.21 vp 4768 4737 4552 96.09 95.47 95.78 sbar 503 489 453 92.64 90.06 91.33 adjp 384 382 320 83.77 83.33 83.55 advp 822 811 709 87.42 86.25 86.83 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.10 80.35 81.22 Avg2. 23486 23473 22523 95.95 95.90 95.93 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 186 Log-likelihood = -18899.049413 Norm (log-likelihood gradient vector) = 602.347990 Norm (lambda vector) = 443.987328 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4579 98.11 98.83 98.47 i-np 13660 13685 13396 97.89 98.07 97.98 e-np 12220 12244 12037 98.31 98.50 98.41 o 6349 6343 6177 97.38 97.29 97.34 e-vp 4768 4734 4662 98.48 97.78 98.13 i-vp 2602 2648 2529 95.51 97.19 96.34 e-adjp 384 378 336 88.89 87.50 88.19 i-pp 52 46 36 78.26 69.23 73.47 e-advp 822 812 717 88.30 87.23 87.76 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 488 459 94.06 91.25 92.63 i-adjp 152 133 115 86.47 75.66 80.70 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.95 74.32 75.61 Avg2. 46451 46451 45289 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12244 11776 96.18 96.37 96.27 pp 4633 4667 4567 97.86 98.58 98.22 vp 4768 4734 4553 96.18 95.49 95.83 sbar 503 488 452 92.62 89.86 91.22 adjp 384 378 317 83.86 82.55 83.20 advp 822 812 711 87.56 86.50 87.03 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.12 80.28 81.19 Avg2. 23486 23477 22520 95.92 95.89 95.91 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 187 Log-likelihood = -18595.943989 Norm (log-likelihood gradient vector) = 733.542995 Norm (lambda vector) = 447.872634 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4579 98.14 98.83 98.48 i-np 13660 13729 13398 97.59 98.08 97.83 e-np 12220 12221 12020 98.36 98.36 98.36 o 6349 6316 6158 97.50 96.99 97.24 e-vp 4768 4733 4663 98.52 97.80 98.16 i-vp 2602 2650 2532 95.55 97.31 96.42 e-adjp 384 380 337 88.68 87.76 88.22 i-pp 52 47 36 76.60 69.23 72.73 e-advp 822 812 717 88.30 87.23 87.76 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 488 459 94.06 91.25 92.63 i-adjp 152 135 115 85.19 75.66 80.14 e-prt 126 132 124 93.94 98.41 96.12 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.61 73.80 75.18 Avg2. 46451 46451 45258 97.43 97.43 97.43 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12221 11748 96.13 96.14 96.13 pp 4633 4666 4568 97.90 98.60 98.25 vp 4768 4733 4557 96.28 95.57 95.93 sbar 503 488 452 92.62 89.86 91.22 adjp 384 380 317 83.42 82.55 82.98 advp 822 812 711 87.56 86.50 87.03 prt 126 132 124 93.94 98.41 96.12 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.95 79.64 80.78 Avg2. 23486 23454 22496 95.92 95.78 95.85 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 188 Log-likelihood = -18421.335432 Norm (log-likelihood gradient vector) = 1641.562415 Norm (lambda vector) = 455.999585 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 320 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4578 98.11 98.81 98.46 i-np 13660 13712 13398 97.71 98.08 97.90 e-np 12220 12227 12028 98.37 98.43 98.40 o 6349 6327 6170 97.52 97.18 97.35 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2649 2530 95.51 97.23 96.36 e-adjp 384 380 337 88.68 87.76 88.22 i-pp 52 47 36 76.60 69.23 72.73 e-advp 822 812 717 88.30 87.23 87.76 i-advp 100 86 72 83.72 72.00 77.42 e-sbar 503 490 460 93.88 91.45 92.65 i-adjp 152 135 115 85.19 75.66 80.14 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 21 20 95.24 83.33 88.89 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.64 73.82 75.20 Avg2. 46451 46451 45274 97.47 97.47 97.47 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12227 11761 96.19 96.24 96.22 pp 4633 4666 4566 97.86 98.55 98.20 vp 4768 4733 4554 96.22 95.51 95.86 sbar 503 490 453 92.45 90.06 91.24 adjp 384 380 317 83.42 82.55 82.98 advp 822 812 711 87.56 86.50 87.03 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.00 79.66 80.81 Avg2. 23486 23461 22505 95.93 95.82 95.87 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 189 Log-likelihood = -18355.317791 Norm (log-likelihood gradient vector) = 930.640631 Norm (lambda vector) = 452.106496 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4579 98.16 98.83 98.49 i-np 13660 13705 13408 97.83 98.16 97.99 e-np 12220 12239 12033 98.32 98.47 98.39 o 6349 6335 6174 97.46 97.24 97.35 e-vp 4768 4730 4662 98.56 97.78 98.17 i-vp 2602 2649 2532 95.58 97.31 96.44 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 45 36 80.00 69.23 74.23 e-advp 822 809 716 88.50 87.10 87.80 i-advp 100 85 72 84.71 72.00 77.84 e-sbar 503 487 458 94.05 91.05 92.53 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.98 74.33 75.63 Avg2. 46451 46451 45296 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12239 11772 96.18 96.33 96.26 pp 4633 4665 4568 97.92 98.60 98.26 vp 4768 4730 4556 96.32 95.55 95.94 sbar 503 487 451 92.61 89.66 91.11 adjp 384 379 317 83.64 82.55 83.09 advp 822 809 710 87.76 86.37 87.06 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.14 80.25 81.18 Avg2. 23486 23463 22518 95.97 95.88 95.93 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 190 Log-likelihood = -18087.083509 Norm (log-likelihood gradient vector) = 422.109084 Norm (lambda vector) = 454.028668 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4580 98.14 98.86 98.49 i-np 13660 13706 13409 97.83 98.16 98.00 e-np 12220 12236 12033 98.34 98.47 98.41 o 6349 6337 6175 97.44 97.26 97.35 e-vp 4768 4728 4662 98.60 97.78 98.19 i-vp 2602 2650 2534 95.62 97.39 96.50 e-adjp 384 380 337 88.68 87.76 88.22 i-pp 52 44 35 79.55 67.31 72.92 e-advp 822 809 717 88.63 87.23 87.92 i-advp 100 83 72 86.75 72.00 78.69 e-sbar 503 488 459 94.06 91.25 92.63 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.06 74.25 75.63 Avg2. 46451 46451 45302 97.53 97.53 97.53 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12236 11773 96.22 96.34 96.28 pp 4633 4667 4568 97.88 98.60 98.24 vp 4768 4728 4557 96.38 95.57 95.98 sbar 503 488 452 92.62 89.86 91.22 adjp 384 380 318 83.68 82.81 83.25 advp 822 809 711 87.89 86.50 87.19 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.16 80.31 81.23 Avg2. 23486 23462 22523 96.00 95.90 95.95 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 191 Log-likelihood = -17963.335949 Norm (log-likelihood gradient vector) = 409.747058 Norm (lambda vector) = 454.136061 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4670 4582 98.12 98.90 98.51 i-np 13660 13713 13411 97.80 98.18 97.99 e-np 12220 12236 12033 98.34 98.47 98.41 o 6349 6333 6172 97.46 97.21 97.33 e-vp 4768 4727 4661 98.60 97.76 98.18 i-vp 2602 2647 2533 95.69 97.35 96.51 e-adjp 384 381 338 88.71 88.02 88.37 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 809 717 88.63 87.23 87.92 i-advp 100 83 72 86.75 72.00 78.69 e-sbar 503 484 457 94.42 90.85 92.60 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.99 74.24 75.59 Avg2. 46451 46451 45300 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12236 11772 96.21 96.33 96.27 pp 4633 4670 4569 97.84 98.62 98.23 vp 4768 4727 4557 96.40 95.57 95.99 sbar 503 484 450 92.98 89.46 91.19 adjp 384 381 319 83.73 83.07 83.40 advp 822 809 711 87.89 86.50 87.19 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.20 80.30 81.24 Avg2. 23486 23461 22522 96.00 95.90 95.95 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 192 Log-likelihood = -17767.116018 Norm (log-likelihood gradient vector) = 502.647057 Norm (lambda vector) = 454.986701 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4664 4581 98.22 98.88 98.55 i-np 13660 13760 13420 97.53 98.24 97.88 e-np 12220 12209 12016 98.42 98.33 98.37 o 6349 6312 6156 97.53 96.96 97.24 e-vp 4768 4726 4659 98.58 97.71 98.15 i-vp 2602 2653 2535 95.55 97.43 96.48 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 808 717 88.74 87.23 87.98 i-advp 100 83 72 86.75 72.00 78.69 e-sbar 503 488 461 94.47 91.65 93.04 i-adjp 152 135 116 85.93 76.32 80.84 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.00 74.29 75.62 Avg2. 46451 46451 45279 97.48 97.48 97.48 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12209 11754 96.27 96.19 96.23 pp 4633 4664 4568 97.94 98.60 98.27 vp 4768 4726 4555 96.38 95.53 95.96 sbar 503 488 454 93.03 90.26 91.62 adjp 384 379 319 84.17 83.07 83.62 advp 822 808 711 88.00 86.50 87.24 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.28 80.36 81.30 Avg2. 23486 23428 22505 96.06 95.82 95.94 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 193 Log-likelihood = -17629.227220 Norm (log-likelihood gradient vector) = 1523.318782 Norm (lambda vector) = 458.416222 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4665 4581 98.20 98.88 98.54 i-np 13660 13709 13411 97.83 98.18 98.00 e-np 12220 12236 12038 98.38 98.51 98.45 o 6349 6335 6177 97.51 97.29 97.40 e-vp 4768 4727 4660 98.58 97.73 98.16 i-vp 2602 2651 2534 95.59 97.39 96.48 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 45 35 77.78 67.31 72.16 e-advp 822 810 717 88.52 87.23 87.87 i-advp 100 83 72 86.75 72.00 78.69 e-sbar 503 487 460 94.46 91.45 92.93 i-adjp 152 135 116 85.93 76.32 80.84 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.00 74.30 75.63 Avg2. 46451 46451 45312 97.55 97.55 97.55 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12236 11784 96.31 96.43 96.37 pp 4633 4665 4568 97.92 98.60 98.26 vp 4768 4727 4556 96.38 95.55 95.97 sbar 503 487 453 93.02 90.06 91.52 adjp 384 379 319 84.17 83.07 83.62 advp 822 810 711 87.78 86.50 87.13 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.25 80.36 81.30 Avg2. 23486 23458 22535 96.07 95.95 96.01 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 194 Log-likelihood = -17445.547916 Norm (log-likelihood gradient vector) = 627.641420 Norm (lambda vector) = 459.482561 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4581 98.16 98.88 98.52 i-np 13660 13711 13409 97.80 98.16 97.98 e-np 12220 12235 12036 98.37 98.49 98.43 o 6349 6333 6176 97.52 97.28 97.40 e-vp 4768 4727 4661 98.60 97.76 98.18 i-vp 2602 2651 2535 95.62 97.43 96.52 e-adjp 384 380 338 88.95 88.02 88.48 i-pp 52 48 35 72.92 67.31 70.00 e-advp 822 808 716 88.61 87.10 87.85 i-advp 100 83 72 86.75 72.00 78.69 e-sbar 503 487 460 94.46 91.45 92.93 i-adjp 152 135 116 85.93 76.32 80.84 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 20 19 95.00 79.17 86.36 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.71 73.58 75.11 Avg2. 46451 46451 45306 97.54 97.54 97.54 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12235 11777 96.26 96.37 96.32 pp 4633 4667 4568 97.88 98.60 98.24 vp 4768 4727 4558 96.42 95.60 96.01 sbar 503 487 453 93.02 90.06 91.52 adjp 384 380 320 84.21 83.33 83.77 advp 822 808 710 87.87 86.37 87.12 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.20 79.75 80.96 Avg2. 23486 23457 22529 96.04 95.93 95.98 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 195 Log-likelihood = -17363.298142 Norm (log-likelihood gradient vector) = 419.359165 Norm (lambda vector) = 460.131175 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4667 4580 98.14 98.86 98.49 i-np 13660 13713 13406 97.76 98.14 97.95 e-np 12220 12231 12033 98.38 98.47 98.43 o 6349 6331 6174 97.52 97.24 97.38 e-vp 4768 4731 4662 98.54 97.78 98.16 i-vp 2602 2646 2533 95.73 97.35 96.53 e-adjp 384 380 337 88.68 87.76 88.22 i-pp 52 48 35 72.92 67.31 70.00 e-advp 822 814 719 88.33 87.47 87.90 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 487 460 94.46 91.45 92.93 i-adjp 152 133 114 85.71 75.00 80.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 20 19 95.00 79.17 86.36 e-conjp 16 12 11 91.67 68.75 78.57 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.62 73.51 75.03 Avg2. 46451 46451 45296 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12231 11769 96.22 96.31 96.27 pp 4633 4667 4567 97.86 98.58 98.22 vp 4768 4731 4559 96.36 95.62 95.99 sbar 503 487 453 93.02 90.06 91.52 adjp 384 380 318 83.68 82.81 83.25 advp 822 814 713 87.59 86.74 87.16 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 12 11 91.67 68.75 78.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.11 79.73 80.90 Avg2. 23486 23463 22522 95.99 95.90 95.94 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 196 Log-likelihood = -17198.666079 Norm (log-likelihood gradient vector) = 617.323808 Norm (lambda vector) = 463.719273 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4671 4582 98.09 98.90 98.50 i-np 13660 13734 13415 97.68 98.21 97.94 e-np 12220 12219 12027 98.43 98.42 98.42 o 6349 6319 6169 97.63 97.16 97.40 e-vp 4768 4730 4662 98.56 97.78 98.17 i-vp 2602 2644 2532 95.76 97.31 96.53 e-adjp 384 379 337 88.92 87.76 88.34 i-pp 52 49 36 73.47 69.23 71.29 e-advp 822 814 719 88.33 87.47 87.90 i-advp 100 84 72 85.71 72.00 78.26 e-sbar 503 485 460 94.85 91.45 93.12 i-adjp 152 133 114 85.71 75.00 80.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 14 10 71.43 83.33 76.92 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.85 74.75 75.78 Avg2. 46451 46451 45300 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12219 11763 96.27 96.26 96.26 pp 4633 4671 4570 97.84 98.64 98.24 vp 4768 4730 4559 96.38 95.62 96.00 sbar 503 485 454 93.61 90.26 91.90 adjp 384 379 319 84.17 83.07 83.62 advp 822 814 713 87.59 86.74 87.16 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.28 80.40 81.33 Avg2. 23486 23452 22522 96.03 95.90 95.96 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 197 Log-likelihood = -16991.388331 Norm (log-likelihood gradient vector) = 617.209774 Norm (lambda vector) = 467.526873 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4669 4580 98.09 98.86 98.47 i-np 13660 13727 13414 97.72 98.20 97.96 e-np 12220 12222 12028 98.41 98.43 98.42 o 6349 6322 6171 97.61 97.20 97.40 e-vp 4768 4731 4662 98.54 97.78 98.16 i-vp 2602 2645 2533 95.77 97.35 96.55 e-adjp 384 381 338 88.71 88.02 88.37 i-pp 52 49 36 73.47 69.23 71.29 e-advp 822 816 719 88.11 87.47 87.79 i-advp 100 83 72 86.75 72.00 78.69 e-sbar 503 484 459 94.83 91.25 93.01 i-adjp 152 133 114 85.71 75.00 80.00 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.77 74.33 75.53 Avg2. 46451 46451 45300 97.52 97.52 97.52 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12222 11765 96.26 96.28 96.27 pp 4633 4669 4568 97.84 98.60 98.22 vp 4768 4731 4558 96.34 95.60 95.97 sbar 503 484 452 93.39 89.86 91.59 adjp 384 381 319 83.73 83.07 83.40 advp 822 816 713 87.38 86.74 87.06 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.19 80.36 81.26 Avg2. 23486 23457 22519 96.00 95.88 95.94 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 198 Log-likelihood = -16814.885132 Norm (log-likelihood gradient vector) = 504.269163 Norm (lambda vector) = 471.113142 Log-likelihood and gradient computational time: 320 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4672 4581 98.05 98.88 98.46 i-np 13660 13730 13411 97.68 98.18 97.93 e-np 12220 12225 12026 98.37 98.41 98.39 o 6349 6318 6167 97.61 97.13 97.37 e-vp 4768 4730 4660 98.52 97.73 98.13 i-vp 2602 2648 2534 95.69 97.39 96.53 e-adjp 384 383 339 88.51 88.28 88.40 i-pp 52 49 36 73.47 69.23 71.29 e-advp 822 811 718 88.53 87.35 87.94 i-advp 100 81 72 88.89 72.00 79.56 e-sbar 503 481 456 94.80 90.66 92.68 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.88 74.34 75.59 Avg2. 46451 46451 45289 97.50 97.50 97.50 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12225 11760 96.20 96.24 96.22 pp 4633 4672 4568 97.77 98.60 98.18 vp 4768 4730 4555 96.30 95.53 95.91 sbar 503 481 449 93.35 89.26 91.26 adjp 384 383 320 83.55 83.33 83.44 advp 822 811 712 87.79 86.62 87.20 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.19 80.30 81.23 Avg2. 23486 23456 22508 95.96 95.84 95.90 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds Iteration: 199 Log-likelihood = -16584.553063 Norm (log-likelihood gradient vector) = 580.745187 Norm (lambda vector) = 476.886147 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 322 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4657 4574 98.22 98.73 98.47 i-np 13660 13687 13390 97.83 98.02 97.93 e-np 12220 12247 12032 98.24 98.46 98.35 o 6349 6348 6180 97.35 97.34 97.35 e-vp 4768 4733 4661 98.48 97.76 98.12 i-vp 2602 2649 2533 95.62 97.35 96.48 e-adjp 384 378 336 88.89 87.50 88.19 i-pp 52 46 35 76.09 67.31 71.43 e-advp 822 809 715 88.38 86.98 87.68 i-advp 100 79 70 88.61 70.00 78.21 e-sbar 503 495 463 93.54 92.05 92.79 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.94 74.15 75.52 Avg2. 46451 46451 45278 97.47 97.47 97.47 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12247 11763 96.05 96.26 96.15 pp 4633 4657 4561 97.94 98.45 98.19 vp 4768 4733 4554 96.22 95.51 95.86 sbar 503 495 456 92.12 90.66 91.38 adjp 384 378 317 83.86 82.55 83.20 advp 822 809 709 87.64 86.25 86.94 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.08 80.31 81.18 Avg2. 23486 23473 22504 95.87 95.82 95.85 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 356 seconds Iteration: 200 Log-likelihood = -16562.954433 Norm (log-likelihood gradient vector) = 1447.819667 Norm (lambda vector) = 485.162641 Log-likelihood and gradient computational time: 321 seconds Training iteration elapsed: 321 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-pp 4633 4666 4579 98.14 98.83 98.48 i-np 13660 13715 13408 97.76 98.16 97.96 e-np 12220 12238 12032 98.32 98.46 98.39 o 6349 6325 6173 97.60 97.23 97.41 e-vp 4768 4731 4661 98.52 97.76 98.14 i-vp 2602 2650 2534 95.62 97.39 96.50 e-adjp 384 381 338 88.71 88.02 88.37 i-pp 52 48 35 72.92 67.31 70.00 e-advp 822 808 715 88.49 86.98 87.73 i-advp 100 80 71 88.75 71.00 78.89 e-sbar 503 486 460 94.65 91.45 93.02 i-adjp 152 134 115 85.82 75.66 80.42 e-prt 126 131 124 94.66 98.41 96.50 i-sbar 12 13 9 69.23 75.00 72.00 i-conjp 24 22 21 95.45 87.50 91.30 e-conjp 16 13 12 92.31 75.00 82.76 e-intj 4 0 0 0.00 0.00 0.00 e-lst 10 10 8 80.00 80.00 80.00 i-ucp 0 0 0 0.00 0.00 0.00 e-ucp 0 0 0 0.00 0.00 0.00 i-intj 12 0 0 0.00 0.00 0.00 i-prt 0 0 0 0.00 0.00 0.00 i-lst 2 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.85 74.21 75.50 Avg2. 46451 46451 45295 97.51 97.51 97.51 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12220 12238 11769 96.17 96.31 96.24 pp 4633 4666 4565 97.84 98.53 98.18 vp 4768 4731 4555 96.28 95.53 95.90 sbar 503 486 453 93.21 90.06 91.61 adjp 384 381 319 83.73 83.07 83.40 advp 822 808 709 87.75 86.25 86.99 prt 126 131 124 94.66 98.41 96.50 lst 10 10 8 80.00 80.00 80.00 intj 4 0 0 0.00 0.00 0.00 conjp 16 13 12 92.31 75.00 82.76 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.19 80.32 81.24 Avg2. 23486 23464 22514 95.95 95.86 95.91 Current max chunk-based F1: 96.15 (iteration 124) Training iteration elapsed (including evaluation time): 355 seconds The training process elapsed: 71108 seconds