OPTION VALUES: Model directory: ./IOE2+0.05/ 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: 23 Number of training sequences: 100 (one data partition) Number of testing sequences: 23 (one data partition) Number of unlabeled sequences: 0 Number of context predicates: 213369 Number of features: 450063 Feature rare threshold: 1 Context predicate rare threshold: 1 Using multiple rare thresholds for features: 0 Highlight feature: 0 Number of training iterations: 130 Initial lambda value: 0.0500 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 = -1055851.589185 Norm (log-likelihood gradient vector) = 116501.020857 Norm (lambda vector) = 33.543367 Log-likelihood and gradient computational time: 74 seconds Training iteration elapsed: 75 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12533 11636 92.84 93.67 93.26 e-pp 4811 5104 4597 90.07 95.55 92.73 i-np 14376 16649 14044 84.35 97.69 90.53 i-vp 2646 2343 2245 95.82 84.85 90.00 e-vp 4658 4132 3893 94.22 83.58 88.58 e-sbar 535 180 176 97.78 32.90 49.23 o 6180 5812 5604 96.42 90.68 93.46 e-adjp 438 122 120 98.36 27.40 42.86 i-advp 89 6 5 83.33 5.62 10.53 e-advp 866 442 401 90.72 46.30 61.31 i-adjp 167 10 10 100.00 5.99 11.30 i-sbar 4 3 1 33.33 25.00 28.57 i-pp 48 15 14 93.33 29.17 44.44 e-prt 106 24 20 83.33 18.87 30.77 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 68.55 40.96 51.28 Avg2. 47375 47375 42766 90.27 90.27 90.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12546 10121 80.67 81.48 81.07 pp 4811 5104 4581 89.75 95.22 92.41 vp 4658 4133 3756 90.88 80.64 85.45 sbar 535 180 173 96.11 32.34 48.39 adjp 438 122 119 97.54 27.17 42.50 advp 866 442 398 90.05 45.96 60.86 prt 106 24 20 83.33 18.87 30.77 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 62.83 38.17 47.49 Avg2. 23852 22551 19168 85.00 80.36 82.62 Current max chunk-based F1: 82.62 (iteration 1) Training iteration elapsed (including evaluation time): 95 seconds Iteration: 2 Log-likelihood = -941597.831951 Norm (log-likelihood gradient vector) = 111630.189494 Norm (lambda vector) = 33.624037 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12223 9818 80.32 79.04 79.68 e-pp 4811 5314 4081 76.80 84.83 80.61 i-np 14376 22294 13950 62.57 97.04 76.08 i-vp 2646 771 763 98.96 28.84 44.66 e-vp 4658 1653 1311 79.31 28.15 41.55 e-sbar 535 0 0 0.00 0.00 0.00 o 6180 5120 4692 91.64 75.92 83.04 e-adjp 438 0 0 0.00 0.00 0.00 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 0 0 0.00 0.00 0.00 i-adjp 167 0 0 0.00 0.00 0.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 27.20 21.88 24.25 Avg2. 47375 47375 34615 73.07 73.07 73.07 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12226 6440 52.67 51.84 52.26 pp 4811 5314 4057 76.35 84.33 80.14 vp 4658 1653 1224 74.05 26.28 38.79 sbar 535 0 0 0.00 0.00 0.00 adjp 438 0 0 0.00 0.00 0.00 advp 866 0 0 0.00 0.00 0.00 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 20.31 16.24 18.05 Avg2. 23852 19193 11721 61.07 49.14 54.46 Current max chunk-based F1: 82.62 (iteration 1) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 3 Log-likelihood = -575903.268055 Norm (log-likelihood gradient vector) = 69866.243264 Norm (lambda vector) = 34.238127 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 11143 8591 77.10 69.16 72.91 e-pp 4811 5184 4069 78.49 84.58 81.42 i-np 14376 23814 13996 58.77 97.36 73.30 i-vp 2646 678 667 98.38 25.21 40.13 e-vp 4658 1469 1234 84.00 26.49 40.28 e-sbar 535 0 0 0.00 0.00 0.00 o 6180 5087 4674 91.88 75.63 82.97 e-adjp 438 0 0 0.00 0.00 0.00 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 0 0 0.00 0.00 0.00 i-adjp 167 0 0 0.00 0.00 0.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 27.15 21.02 23.70 Avg2. 47375 47375 33231 70.14 70.14 70.14 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 11147 5389 48.34 43.38 45.73 pp 4811 5184 4046 78.05 84.10 80.96 vp 4658 1469 1070 72.84 22.97 34.93 sbar 535 0 0 0.00 0.00 0.00 adjp 438 0 0 0.00 0.00 0.00 advp 866 0 0 0.00 0.00 0.00 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 19.92 15.05 17.14 Avg2. 23852 17800 10505 59.02 44.04 50.44 Current max chunk-based F1: 82.62 (iteration 1) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 4 Log-likelihood = -414435.344645 Norm (log-likelihood gradient vector) = 76621.112011 Norm (lambda vector) = 37.424540 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 13250 11002 83.03 88.57 85.71 e-pp 4811 5128 4372 85.26 90.88 87.98 i-np 14376 17142 13588 79.27 94.52 86.22 i-vp 2646 1884 1772 94.06 66.97 78.23 e-vp 4658 4010 3367 83.97 72.28 77.69 e-sbar 535 0 0 0.00 0.00 0.00 o 6180 5961 5429 91.08 87.85 89.43 e-adjp 438 0 0 0.00 0.00 0.00 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 0 0 0.00 0.00 0.00 i-adjp 167 0 0 0.00 0.00 0.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 28.70 27.84 28.26 Avg2. 47375 47375 39530 83.44 83.44 83.44 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 13252 8935 67.42 71.93 69.60 pp 4811 5128 4342 84.67 90.25 87.37 vp 4658 4010 3149 78.53 67.60 72.66 sbar 535 0 0 0.00 0.00 0.00 adjp 438 0 0 0.00 0.00 0.00 advp 866 0 0 0.00 0.00 0.00 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 23.06 22.98 23.02 Avg2. 23852 22390 16426 73.36 68.87 71.04 Current max chunk-based F1: 82.62 (iteration 1) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 5 Log-likelihood = -309877.828784 Norm (log-likelihood gradient vector) = 37357.827390 Norm (lambda vector) = 37.745829 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 13095 11448 87.42 92.16 89.73 e-pp 4811 5300 4514 85.17 93.83 89.29 i-np 14376 15858 13620 85.89 94.74 90.10 i-vp 2646 2658 2264 85.18 85.56 85.37 e-vp 4658 4670 3902 83.55 83.77 83.66 e-sbar 535 0 0 0.00 0.00 0.00 o 6180 5794 5483 94.63 88.72 91.58 e-adjp 438 0 0 0.00 0.00 0.00 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 0 0 0.00 0.00 0.00 i-adjp 167 0 0 0.00 0.00 0.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 28.99 29.93 29.45 Avg2. 47375 47375 41231 87.03 87.03 87.03 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 13101 9976 76.15 80.31 78.17 pp 4811 5300 4482 84.57 93.16 88.66 vp 4658 4670 3637 77.88 78.08 77.98 sbar 535 0 0 0.00 0.00 0.00 adjp 438 0 0 0.00 0.00 0.00 advp 866 0 0 0.00 0.00 0.00 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 23.86 25.16 24.49 Avg2. 23852 23071 18095 78.43 75.86 77.13 Current max chunk-based F1: 82.62 (iteration 1) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 6 Log-likelihood = -257538.898912 Norm (log-likelihood gradient vector) = 19452.505103 Norm (lambda vector) = 37.946003 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12962 11715 90.38 94.31 92.30 e-pp 4811 5347 4565 85.37 94.89 89.88 i-np 14376 15122 13631 90.14 94.82 92.42 i-vp 2646 3230 2472 76.53 93.42 84.14 e-vp 4658 4857 4023 82.83 86.37 84.56 e-sbar 535 0 0 0.00 0.00 0.00 o 6180 5823 5534 95.04 89.55 92.21 e-adjp 438 0 0 0.00 0.00 0.00 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 34 33 97.06 3.81 7.33 i-adjp 167 0 0 0.00 0.00 0.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 34.30 30.95 32.54 Avg2. 47375 47375 41973 88.60 88.60 88.60 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12967 10606 81.79 85.38 83.55 pp 4811 5347 4532 84.76 94.20 89.23 vp 4658 4857 3664 75.44 78.66 77.02 sbar 535 0 0 0.00 0.00 0.00 adjp 438 0 0 0.00 0.00 0.00 advp 866 34 33 97.06 3.81 7.33 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 33.90 26.21 29.56 Avg2. 23852 23205 18835 81.17 78.97 80.05 Current max chunk-based F1: 82.62 (iteration 1) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 7 Log-likelihood = -225178.986815 Norm (log-likelihood gradient vector) = 16255.999014 Norm (lambda vector) = 38.628873 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12640 11869 93.90 95.55 94.72 e-pp 4811 5141 4585 89.18 95.30 92.14 i-np 14376 14762 13734 93.04 95.53 94.27 i-vp 2646 3460 2573 74.36 97.24 84.28 e-vp 4658 4980 4222 84.78 90.64 87.61 e-sbar 535 162 154 95.06 28.79 44.19 o 6180 5810 5568 95.83 90.10 92.88 e-adjp 438 52 50 96.15 11.42 20.41 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 368 310 84.24 35.80 50.24 i-adjp 167 0 0 0.00 0.00 0.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 44.81 35.58 39.66 Avg2. 47375 47375 43065 90.90 90.90 90.90 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12644 11042 87.33 88.89 88.10 pp 4811 5141 4551 88.52 94.60 91.46 vp 4658 4980 3829 76.89 82.20 79.46 sbar 535 162 152 93.83 28.41 43.62 adjp 438 52 50 96.15 11.42 20.41 advp 866 368 290 78.80 33.49 47.00 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 52.15 33.90 41.09 Avg2. 23852 23347 19914 85.30 83.49 84.38 Current max chunk-based F1: 84.38 (iteration 7) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 8 Log-likelihood = -179923.818133 Norm (log-likelihood gradient vector) = 16727.535465 Norm (lambda vector) = 41.230545 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12766 11991 93.93 96.53 95.21 e-pp 4811 5052 4608 91.21 95.78 93.44 i-np 14376 14191 13499 95.12 93.90 94.51 i-vp 2646 3139 2557 81.46 96.64 88.40 e-vp 4658 4831 4293 88.86 92.16 90.48 e-sbar 535 301 280 93.02 52.34 66.99 o 6180 6263 5860 93.57 94.82 94.19 e-adjp 438 200 176 88.00 40.18 55.17 i-advp 89 0 0 0.00 0.00 0.00 e-advp 866 627 483 77.03 55.77 64.70 i-adjp 167 5 4 80.00 2.40 4.65 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 0 0 0.00 0.00 0.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 49.01 40.03 44.07 Avg2. 47375 47375 43751 92.35 92.35 92.35 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12768 11205 87.76 90.20 88.96 pp 4811 5052 4573 90.52 95.05 92.73 vp 4658 4832 4015 83.09 86.20 84.62 sbar 535 301 278 92.36 51.96 66.51 adjp 438 200 137 68.50 31.28 42.95 advp 866 627 452 72.09 52.19 60.55 prt 106 0 0 0.00 0.00 0.00 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 49.43 40.69 44.64 Avg2. 23852 23780 20660 86.88 86.62 86.75 Current max chunk-based F1: 86.75 (iteration 8) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 9 Log-likelihood = -150417.226994 Norm (log-likelihood gradient vector) = 11107.983751 Norm (lambda vector) = 44.364862 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12480 11898 95.34 95.78 95.56 e-pp 4811 5100 4644 91.06 96.53 93.71 i-np 14376 14616 13739 94.00 95.57 94.78 i-vp 2646 2908 2545 87.52 96.18 91.65 e-vp 4658 4798 4358 90.83 93.56 92.17 e-sbar 535 317 292 92.11 54.58 68.54 o 6180 6092 5817 95.49 94.13 94.80 e-adjp 438 287 241 83.97 55.02 66.48 i-advp 89 1 1 100.00 1.12 2.22 e-advp 866 736 550 74.73 63.51 68.66 i-adjp 167 39 29 74.36 17.37 28.16 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 1 1 100.00 0.94 1.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 59.97 42.46 49.72 Avg2. 47375 47375 44115 93.12 93.12 93.12 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12488 11145 89.25 89.72 89.48 pp 4811 5100 4609 90.37 95.80 93.01 vp 4658 4798 4155 86.60 89.20 87.88 sbar 535 317 290 91.48 54.21 68.08 adjp 438 287 193 67.25 44.06 53.24 advp 866 736 517 70.24 59.70 64.54 prt 106 1 1 100.00 0.94 1.87 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 59.52 43.36 50.17 Avg2. 23852 23727 20910 88.13 87.67 87.90 Current max chunk-based F1: 87.90 (iteration 9) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 10 Log-likelihood = -132188.558749 Norm (log-likelihood gradient vector) = 7165.412158 Norm (lambda vector) = 47.219413 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12501 11967 95.73 96.34 96.03 e-pp 4811 5047 4666 92.45 96.99 94.66 i-np 14376 14423 13708 95.04 95.35 95.20 i-vp 2646 2795 2526 90.38 95.46 92.85 e-vp 4658 4776 4413 92.40 94.74 93.56 e-sbar 535 388 353 90.98 65.98 76.49 o 6180 6160 5878 95.42 95.11 95.27 e-adjp 438 373 288 77.21 65.75 71.02 i-advp 89 4 4 100.00 4.49 8.60 e-advp 866 834 620 74.34 71.59 72.94 i-adjp 167 68 47 69.12 28.14 40.00 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 0 0 0.00 0.00 0.00 e-prt 106 6 4 66.67 3.77 7.14 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 57.76 45.21 50.72 Avg2. 47375 47375 44474 93.88 93.88 93.88 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12511 11267 90.06 90.70 90.38 pp 4811 5047 4631 91.76 96.26 93.95 vp 4658 4776 4245 88.88 91.13 89.99 sbar 535 388 351 90.46 65.61 76.06 adjp 438 373 238 63.81 54.34 58.69 advp 866 834 588 70.50 67.90 69.18 prt 106 6 4 66.67 3.77 7.14 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 56.21 46.97 51.18 Avg2. 23852 23935 21324 89.09 89.40 89.25 Current max chunk-based F1: 89.25 (iteration 10) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 11 Log-likelihood = -119182.530701 Norm (log-likelihood gradient vector) = 5745.517306 Norm (lambda vector) = 51.368849 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12314 11893 96.58 95.74 96.16 e-pp 4811 5084 4698 92.41 97.65 94.96 i-np 14376 14619 13828 94.59 96.19 95.38 i-vp 2646 2739 2508 91.57 94.78 93.15 e-vp 4658 4759 4420 92.88 94.89 93.87 e-sbar 535 395 353 89.37 65.98 75.91 o 6180 6109 5853 95.81 94.71 95.26 e-adjp 438 405 304 75.06 69.41 72.12 i-advp 89 16 16 100.00 17.98 30.48 e-advp 866 797 629 78.92 72.63 75.65 i-adjp 167 103 71 68.93 42.51 52.59 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 7 6 85.71 12.50 21.82 e-prt 106 28 20 71.43 18.87 29.85 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 62.96 48.55 54.82 Avg2. 47375 47375 44599 94.14 94.14 94.14 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12332 11223 91.01 90.35 90.68 pp 4811 5084 4667 91.80 97.01 94.33 vp 4658 4760 4261 89.52 91.48 90.49 sbar 535 395 350 88.61 65.42 75.27 adjp 438 405 268 66.17 61.19 63.58 advp 866 797 608 76.29 70.21 73.12 prt 106 28 20 71.43 18.87 29.85 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 57.48 49.45 53.17 Avg2. 23852 23801 21397 89.90 89.71 89.80 Current max chunk-based F1: 89.80 (iteration 11) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 12 Log-likelihood = -110003.846685 Norm (log-likelihood gradient vector) = 7648.173812 Norm (lambda vector) = 57.322201 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12529 12012 95.87 96.70 96.28 e-pp 4811 5046 4703 93.20 97.76 95.42 i-np 14376 14255 13663 95.85 95.04 95.44 i-vp 2646 2719 2508 92.24 94.78 93.49 e-vp 4658 4770 4435 92.98 95.21 94.08 e-sbar 535 401 366 91.27 68.41 78.21 o 6180 6254 5929 94.80 95.94 95.37 e-adjp 438 405 313 77.28 71.46 74.26 i-advp 89 31 26 83.87 29.21 43.33 e-advp 866 775 632 81.55 72.98 77.03 i-adjp 167 112 78 69.64 46.71 55.91 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 17 16 94.12 33.33 49.23 e-prt 106 61 45 73.77 42.45 53.89 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.14 52.22 57.16 Avg2. 47375 47375 44726 94.41 94.41 94.41 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12539 11365 90.64 91.49 91.06 pp 4811 5046 4682 92.79 97.32 95.00 vp 4658 4770 4283 89.79 91.95 90.86 sbar 535 401 363 90.52 67.85 77.56 adjp 438 405 280 69.14 63.93 66.43 advp 866 775 615 79.35 71.02 74.95 prt 106 61 45 73.77 42.45 53.89 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 58.60 52.60 55.44 Avg2. 23852 23997 21633 90.15 90.70 90.42 Current max chunk-based F1: 90.42 (iteration 12) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 13 Log-likelihood = -102013.549777 Norm (log-likelihood gradient vector) = 4874.365924 Norm (lambda vector) = 61.299217 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12453 11987 96.26 96.50 96.38 e-pp 4811 5013 4697 93.70 97.63 95.62 i-np 14376 14431 13757 95.33 95.69 95.51 i-vp 2646 2715 2511 92.49 94.90 93.68 e-vp 4658 4746 4429 93.32 95.08 94.19 e-sbar 535 412 377 91.50 70.47 79.62 o 6180 6200 5917 95.44 95.74 95.59 e-adjp 438 392 316 80.61 72.15 76.14 i-advp 89 45 31 68.89 34.83 46.27 e-advp 866 756 624 82.54 72.06 76.94 i-adjp 167 117 83 70.94 49.70 58.45 i-sbar 4 0 0 0.00 0.00 0.00 i-pp 48 26 25 96.15 52.08 67.57 e-prt 106 69 52 75.36 49.06 59.43 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 62.92 54.22 58.24 Avg2. 47375 47375 44806 94.58 94.58 94.58 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12460 11343 91.04 91.31 91.17 pp 4811 5013 4685 93.46 97.38 95.38 vp 4658 4746 4288 90.35 92.06 91.20 sbar 535 412 374 90.78 69.91 78.99 adjp 438 392 284 72.45 64.84 68.43 advp 866 756 609 80.56 70.32 75.09 prt 106 69 52 75.36 49.06 59.43 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 59.40 53.49 56.29 Avg2. 23852 23848 21635 90.72 90.71 90.71 Current max chunk-based F1: 90.71 (iteration 13) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 14 Log-likelihood = -97634.128973 Norm (log-likelihood gradient vector) = 4412.280313 Norm (lambda vector) = 63.545869 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12543 12038 95.97 96.91 96.44 e-pp 4811 4969 4687 94.32 97.42 95.85 i-np 14376 14338 13723 95.71 95.46 95.58 i-vp 2646 2729 2523 92.45 95.35 93.88 e-vp 4658 4725 4430 93.76 95.11 94.43 e-sbar 535 427 394 92.27 73.64 81.91 o 6180 6226 5934 95.31 96.02 95.66 e-adjp 438 371 311 83.83 71.00 76.89 i-advp 89 62 36 58.06 40.45 47.68 e-advp 866 754 621 82.36 71.71 76.67 i-adjp 167 120 88 73.33 52.69 61.32 i-sbar 4 5 1 20.00 25.00 22.22 i-pp 48 31 26 83.87 54.17 65.82 e-prt 106 75 55 73.33 51.89 60.77 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.03 56.49 59.58 Avg2. 47375 47375 44867 94.71 94.71 94.71 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12549 11411 90.93 91.86 91.39 pp 4811 4969 4672 94.02 97.11 95.54 vp 4658 4725 4295 90.90 92.21 91.55 sbar 535 427 388 90.87 72.52 80.67 adjp 438 371 285 76.82 65.07 70.46 advp 866 754 604 80.11 69.75 74.57 prt 106 75 55 73.33 51.89 60.77 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 59.70 54.04 56.73 Avg2. 23852 23870 21710 90.95 91.02 90.99 Current max chunk-based F1: 90.99 (iteration 14) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 15 Log-likelihood = -92943.972245 Norm (log-likelihood gradient vector) = 5801.366738 Norm (lambda vector) = 68.147704 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12428 11991 96.48 96.53 96.51 e-pp 4811 4922 4675 94.98 97.17 96.06 i-np 14376 14578 13865 95.11 96.45 95.77 i-vp 2646 2753 2536 92.12 95.84 93.94 e-vp 4658 4700 4422 94.09 94.93 94.51 e-sbar 535 462 414 89.61 77.38 83.05 o 6180 6094 5889 96.64 95.29 95.96 e-adjp 438 347 299 86.17 68.26 76.18 i-advp 89 74 40 54.05 44.94 49.08 e-advp 866 790 641 81.14 74.02 77.42 i-adjp 167 108 86 79.63 51.50 62.55 i-sbar 4 9 1 11.11 25.00 15.38 i-pp 48 36 27 75.00 56.25 64.29 e-prt 106 74 58 78.38 54.72 64.44 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 62.47 57.13 59.68 Avg2. 47375 47375 44944 94.87 94.87 94.87 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12432 11393 91.64 91.72 91.68 pp 4811 4922 4658 94.64 96.82 95.72 vp 4658 4700 4287 91.21 92.04 91.62 sbar 535 462 404 87.45 75.51 81.04 adjp 438 348 278 79.89 63.47 70.74 advp 866 790 624 78.99 72.06 75.36 prt 106 74 58 78.38 54.72 64.44 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 60.22 54.63 57.29 Avg2. 23852 23728 21702 91.46 90.99 91.22 Current max chunk-based F1: 91.22 (iteration 15) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 16 Log-likelihood = -87855.960711 Norm (log-likelihood gradient vector) = 4115.227240 Norm (lambda vector) = 70.750599 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12433 12005 96.56 96.64 96.60 e-pp 4811 4940 4687 94.88 97.42 96.13 i-np 14376 14560 13879 95.32 96.54 95.93 i-vp 2646 2755 2538 92.12 95.92 93.98 e-vp 4658 4678 4419 94.46 94.87 94.67 e-sbar 535 440 402 91.36 75.14 82.46 o 6180 6107 5907 96.73 95.58 96.15 e-adjp 438 334 295 88.32 67.35 76.42 i-advp 89 79 42 53.16 47.19 50.00 e-advp 866 838 670 79.95 77.37 78.64 i-adjp 167 102 86 84.31 51.50 63.94 i-sbar 4 9 1 11.11 25.00 15.38 i-pp 48 33 26 78.79 54.17 64.20 e-prt 106 67 56 83.58 52.83 64.74 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.37 57.08 60.06 Avg2. 47375 47375 45013 95.01 95.01 95.01 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12436 11430 91.91 92.01 91.96 pp 4811 4940 4670 94.53 97.07 95.79 vp 4658 4678 4282 91.53 91.93 91.73 sbar 535 440 392 89.09 73.27 80.41 adjp 438 335 278 82.99 63.47 71.93 advp 866 838 651 77.68 75.17 76.41 prt 106 67 56 83.58 52.83 64.74 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 61.13 54.58 57.67 Avg2. 23852 23734 21759 91.68 91.23 91.45 Current max chunk-based F1: 91.45 (iteration 16) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 17 Log-likelihood = -83363.855207 Norm (log-likelihood gradient vector) = 3234.085594 Norm (lambda vector) = 71.423913 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12319 11952 97.02 96.22 96.62 e-pp 4811 4965 4708 94.82 97.86 96.32 i-np 14376 14775 13988 94.67 97.30 95.97 i-vp 2646 2736 2532 92.54 95.69 94.09 e-vp 4658 4663 4424 94.87 94.98 94.93 e-sbar 535 427 395 92.51 73.83 82.12 o 6180 5972 5836 97.72 94.43 96.05 e-adjp 438 354 306 86.44 69.86 77.27 i-advp 89 83 44 53.01 49.44 51.16 e-advp 866 874 686 78.49 79.21 78.85 i-adjp 167 100 86 86.00 51.50 64.42 i-sbar 4 9 1 11.11 25.00 15.38 i-pp 48 28 25 89.29 52.08 65.79 e-prt 106 70 58 82.86 54.72 65.91 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 0 0 0.00 0.00 0.00 e-conjp 9 0 0 0.00 0.00 0.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.96 57.34 60.47 Avg2. 47375 47375 45041 95.07 95.07 95.07 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12320 11386 92.42 91.66 92.04 pp 4811 4965 4693 94.52 97.55 96.01 vp 4658 4663 4291 92.02 92.12 92.07 sbar 535 427 385 90.16 71.96 80.04 adjp 438 355 289 81.41 65.98 72.89 advp 866 874 667 76.32 77.02 76.67 prt 106 70 58 82.86 54.72 65.91 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 0 0 0.00 0.00 0.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 60.97 55.10 57.89 Avg2. 23852 23674 21769 91.95 91.27 91.61 Current max chunk-based F1: 91.61 (iteration 17) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 18 Log-likelihood = -77795.405434 Norm (log-likelihood gradient vector) = 5055.178009 Norm (lambda vector) = 74.340577 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12506 12064 96.47 97.12 96.79 e-pp 4811 4955 4717 95.20 98.05 96.60 i-np 14376 14371 13826 96.21 96.17 96.19 i-vp 2646 2692 2518 93.54 95.16 94.34 e-vp 4658 4669 4440 95.10 95.32 95.21 e-sbar 535 447 413 92.39 77.20 84.11 o 6180 6176 5946 96.28 96.21 96.24 e-adjp 438 387 323 83.46 73.74 78.30 i-advp 89 78 45 57.69 50.56 53.89 e-advp 866 863 693 80.30 80.02 80.16 i-adjp 167 102 87 85.29 52.10 64.68 i-sbar 4 10 1 10.00 25.00 14.29 i-pp 48 28 25 89.29 52.08 65.79 e-prt 106 82 67 81.71 63.21 71.28 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 6 6 100.00 46.15 63.16 e-conjp 9 3 3 100.00 33.33 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.16 62.86 68.46 Avg2. 47375 47375 45174 95.35 95.35 95.35 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12507 11535 92.23 92.86 92.54 pp 4811 4955 4703 94.91 97.76 96.31 vp 4658 4669 4308 92.27 92.49 92.38 sbar 535 447 402 89.93 75.14 81.87 adjp 438 387 304 78.55 69.41 73.70 advp 866 863 675 78.22 77.94 78.08 prt 106 82 67 81.71 63.21 71.28 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 3 3 100.00 33.33 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.78 60.21 65.07 Avg2. 23852 23913 21997 91.99 92.22 92.11 Current max chunk-based F1: 92.11 (iteration 18) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 19 Log-likelihood = -73752.157385 Norm (log-likelihood gradient vector) = 3716.806961 Norm (lambda vector) = 78.468083 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12457 12053 96.76 97.03 96.89 e-pp 4811 4948 4716 95.31 98.03 96.65 i-np 14376 14436 13882 96.16 96.56 96.36 i-vp 2646 2711 2528 93.25 95.54 94.38 e-vp 4658 4672 4441 95.06 95.34 95.20 e-sbar 535 447 413 92.39 77.20 84.11 o 6180 6148 5940 96.62 96.12 96.37 e-adjp 438 399 327 81.95 74.66 78.14 i-advp 89 74 44 59.46 49.44 53.99 e-advp 866 831 680 81.83 78.52 80.14 i-adjp 167 110 91 82.73 54.49 65.70 i-sbar 4 11 1 9.09 25.00 13.33 i-pp 48 28 25 89.29 52.08 65.79 e-prt 106 91 70 76.92 66.04 71.07 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 8 8 100.00 61.54 76.19 e-conjp 9 4 4 100.00 44.44 61.54 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.82 64.56 69.31 Avg2. 47375 47375 45223 95.46 95.46 95.46 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12458 11539 92.62 92.89 92.76 pp 4811 4948 4702 95.03 97.73 96.36 vp 4658 4672 4311 92.27 92.55 92.41 sbar 535 447 401 89.71 74.95 81.67 adjp 438 399 310 77.69 70.78 74.07 advp 866 832 664 79.81 76.67 78.21 prt 106 91 70 76.92 66.04 71.07 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 4 4 100.00 44.44 61.54 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.41 61.61 65.71 Avg2. 23852 23851 22001 92.24 92.24 92.24 Current max chunk-based F1: 92.24 (iteration 19) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 20 Log-likelihood = -71191.505486 Norm (log-likelihood gradient vector) = 2675.153523 Norm (lambda vector) = 79.894720 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12422 12057 97.06 97.06 97.06 e-pp 4811 4907 4708 95.94 97.86 96.89 i-np 14376 14475 13924 96.19 96.86 96.52 i-vp 2646 2716 2531 93.19 95.65 94.41 e-vp 4658 4698 4456 94.85 95.66 95.25 e-sbar 535 480 439 91.46 82.06 86.50 o 6180 6127 5933 96.83 96.00 96.42 e-adjp 438 407 334 82.06 76.26 79.05 i-advp 89 74 44 59.46 49.44 53.99 e-advp 866 782 660 84.40 76.21 80.10 i-adjp 167 124 102 82.26 61.08 70.10 i-sbar 4 13 1 7.69 25.00 11.76 i-pp 48 33 27 81.82 56.25 66.67 e-prt 106 105 76 72.38 71.70 72.04 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 8 8 100.00 61.54 76.19 e-conjp 9 4 4 100.00 44.44 61.54 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.20 65.73 69.71 Avg2. 47375 47375 45304 95.63 95.63 95.63 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12423 11563 93.08 93.08 93.08 pp 4811 4907 4694 95.66 97.57 96.60 vp 4658 4698 4324 92.04 92.83 92.43 sbar 535 480 426 88.75 79.63 83.94 adjp 438 407 317 77.89 72.37 75.03 advp 866 783 645 82.38 74.48 78.23 prt 106 105 76 72.38 71.70 72.04 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 4 4 100.00 44.44 61.54 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.22 62.61 66.20 Avg2. 23852 23807 22049 92.62 92.44 92.53 Current max chunk-based F1: 92.53 (iteration 20) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 21 Log-likelihood = -66538.121153 Norm (log-likelihood gradient vector) = 2100.755715 Norm (lambda vector) = 85.767769 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12302 11985 97.42 96.48 96.95 e-pp 4811 4889 4703 96.20 97.76 96.97 i-np 14376 14705 14011 95.28 97.46 96.36 i-vp 2646 2707 2526 93.31 95.46 94.38 e-vp 4658 4690 4457 95.03 95.68 95.36 e-sbar 535 485 444 91.55 82.99 87.06 o 6180 6076 5898 97.07 95.44 96.25 e-adjp 438 398 336 84.42 76.71 80.38 i-advp 89 73 49 67.12 55.06 60.49 e-advp 866 749 645 86.11 74.48 79.88 i-adjp 167 133 108 81.20 64.67 72.00 i-sbar 4 14 1 7.14 25.00 11.11 i-pp 48 38 31 81.58 64.58 72.09 e-prt 106 104 76 73.08 71.70 72.38 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 8 8 100.00 61.54 76.19 e-conjp 9 4 4 100.00 44.44 61.54 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.81 66.64 70.49 Avg2. 47375 47375 45282 95.58 95.58 95.58 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12306 11478 93.27 92.40 92.83 pp 4811 4889 4693 95.99 97.55 96.76 vp 4658 4690 4324 92.20 92.83 92.51 sbar 535 485 430 88.66 80.37 84.31 adjp 438 398 319 80.15 72.83 76.32 advp 866 750 634 84.53 73.21 78.47 prt 106 104 76 73.08 71.70 72.38 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 4 4 100.00 44.44 61.54 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.79 62.53 66.41 Avg2. 23852 23626 21958 92.94 92.06 92.50 Current max chunk-based F1: 92.53 (iteration 20) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 22 Log-likelihood = -63552.685485 Norm (log-likelihood gradient vector) = 4113.705829 Norm (lambda vector) = 92.010030 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12411 12056 97.14 97.05 97.10 e-pp 4811 4899 4705 96.04 97.80 96.91 i-np 14376 14488 13930 96.15 96.90 96.52 i-vp 2646 2720 2533 93.12 95.73 94.41 e-vp 4658 4687 4457 95.09 95.68 95.39 e-sbar 535 487 447 91.79 83.55 87.48 o 6180 6156 5946 96.59 96.21 96.40 e-adjp 438 395 332 84.05 75.80 79.71 i-advp 89 70 49 70.00 55.06 61.64 e-advp 866 762 653 85.70 75.40 80.22 i-adjp 167 135 110 81.48 65.87 72.85 i-sbar 4 16 2 12.50 50.00 20.00 i-pp 48 38 31 81.58 64.58 72.09 e-prt 106 95 72 75.79 67.92 71.64 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.88 69.45 72.06 Avg2. 47375 47375 45338 95.70 95.70 95.70 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12413 11574 93.24 93.17 93.21 pp 4811 4900 4695 95.82 97.59 96.69 vp 4658 4687 4325 92.28 92.85 92.56 sbar 535 487 433 88.91 80.93 84.74 adjp 438 395 317 80.25 72.37 76.11 advp 866 763 643 84.27 74.25 78.94 prt 106 95 72 75.79 67.92 71.64 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 69.39 63.47 66.30 Avg2. 23852 23746 22064 92.92 92.50 92.71 Current max chunk-based F1: 92.71 (iteration 22) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 23 Log-likelihood = -61249.004324 Norm (log-likelihood gradient vector) = 2125.577034 Norm (lambda vector) = 96.353594 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12424 12063 97.09 97.11 97.10 e-pp 4811 4901 4707 96.04 97.84 96.93 i-np 14376 14471 13923 96.21 96.85 96.53 i-vp 2646 2719 2533 93.16 95.73 94.43 e-vp 4658 4678 4457 95.28 95.68 95.48 e-sbar 535 492 451 91.67 84.30 87.83 o 6180 6157 5949 96.62 96.26 96.44 e-adjp 438 376 321 85.37 73.29 78.87 i-advp 89 74 52 70.27 58.43 63.80 e-advp 866 798 668 83.71 77.14 80.29 i-adjp 167 129 107 82.95 64.07 72.30 i-sbar 4 16 2 12.50 50.00 20.00 i-pp 48 37 30 81.08 62.50 70.59 e-prt 106 87 69 79.31 65.09 71.50 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.12 69.26 72.07 Avg2. 47375 47375 45347 95.72 95.72 95.72 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12427 11592 93.28 93.32 93.30 pp 4811 4902 4696 95.80 97.61 96.70 vp 4658 4678 4326 92.48 92.87 92.67 sbar 535 492 437 88.82 81.68 85.10 adjp 438 376 305 81.12 69.63 74.94 advp 866 799 658 82.35 75.98 79.04 prt 106 87 69 79.31 65.09 71.50 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 69.65 63.17 66.25 Avg2. 23852 23767 22088 92.94 92.60 92.77 Current max chunk-based F1: 92.77 (iteration 23) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 24 Log-likelihood = -59423.276430 Norm (log-likelihood gradient vector) = 1828.004107 Norm (lambda vector) = 100.133150 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12505 12096 96.73 97.38 97.05 e-pp 4811 4905 4712 96.07 97.94 96.99 i-np 14376 14296 13827 96.72 96.18 96.45 i-vp 2646 2704 2536 93.79 95.84 94.80 e-vp 4658 4687 4466 95.28 95.88 95.58 e-sbar 535 494 452 91.50 84.49 87.85 o 6180 6222 5976 96.05 96.70 96.37 e-adjp 438 370 321 86.76 73.29 79.46 i-advp 89 80 54 67.50 60.67 63.91 e-advp 866 841 687 81.69 79.33 80.49 i-adjp 167 130 108 83.08 64.67 72.73 i-sbar 4 14 1 7.14 25.00 11.11 i-pp 48 36 30 83.33 62.50 71.43 e-prt 106 75 61 81.33 57.55 67.40 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.88 67.77 71.15 Avg2. 47375 47375 45342 95.71 95.71 95.71 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12509 11619 92.89 93.54 93.21 pp 4811 4906 4702 95.84 97.73 96.78 vp 4658 4687 4341 92.62 93.19 92.91 sbar 535 494 438 88.66 81.87 85.13 adjp 438 370 305 82.43 69.63 75.50 advp 866 841 677 80.50 78.18 79.32 prt 106 75 61 81.33 57.55 67.40 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 69.76 62.72 66.06 Avg2. 23852 23888 22148 92.72 92.86 92.79 Current max chunk-based F1: 92.79 (iteration 24) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 25 Log-likelihood = -57713.815730 Norm (log-likelihood gradient vector) = 3422.938533 Norm (lambda vector) = 104.752958 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12430 12071 97.11 97.17 97.14 e-pp 4811 4894 4712 96.28 97.94 97.10 i-np 14376 14453 13924 96.34 96.86 96.60 i-vp 2646 2694 2532 93.99 95.69 94.83 e-vp 4658 4682 4462 95.30 95.79 95.55 e-sbar 535 499 456 91.38 85.23 88.20 o 6180 6147 5952 96.83 96.31 96.57 e-adjp 438 371 324 87.33 73.97 80.10 i-advp 89 80 53 66.25 59.55 62.72 e-advp 866 851 697 81.90 80.48 81.19 i-adjp 167 128 108 84.38 64.67 73.22 i-sbar 4 15 2 13.33 50.00 21.05 i-pp 48 36 30 83.33 62.50 71.43 e-prt 106 79 65 82.28 61.32 70.27 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.39 69.44 72.29 Avg2. 47375 47375 45403 95.84 95.84 95.84 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12434 11610 93.37 93.46 93.42 pp 4811 4895 4702 96.06 97.73 96.89 vp 4658 4682 4340 92.70 93.17 92.93 sbar 535 499 443 88.78 82.80 85.69 adjp 438 371 308 83.02 70.32 76.14 advp 866 851 686 80.61 79.21 79.91 prt 106 79 65 82.28 61.32 70.27 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 70.01 63.36 66.52 Avg2. 23852 23817 22159 93.04 92.90 92.97 Current max chunk-based F1: 92.97 (iteration 25) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 26 Log-likelihood = -55449.294708 Norm (log-likelihood gradient vector) = 1831.594982 Norm (lambda vector) = 107.275435 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12404 12062 97.24 97.10 97.17 e-pp 4811 4894 4719 96.42 98.09 97.25 i-np 14376 14497 13953 96.25 97.06 96.65 i-vp 2646 2683 2532 94.37 95.69 95.03 e-vp 4658 4677 4465 95.47 95.86 95.66 e-sbar 535 502 462 92.03 86.36 89.10 o 6180 6126 5942 97.00 96.15 96.57 e-adjp 438 394 336 85.28 76.71 80.77 i-advp 89 76 51 67.11 57.30 61.82 e-advp 866 832 695 83.53 80.25 81.86 i-adjp 167 139 110 79.14 65.87 71.90 i-sbar 4 15 2 13.33 50.00 21.05 i-pp 48 35 30 85.71 62.50 72.29 e-prt 106 86 69 80.23 65.09 71.88 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.73 69.81 72.65 Avg2. 47375 47375 45443 95.92 95.92 95.92 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12406 11604 93.54 93.41 93.48 pp 4811 4894 4709 96.22 97.88 97.04 vp 4658 4677 4351 93.03 93.41 93.22 sbar 535 502 449 89.44 83.93 86.60 adjp 438 394 319 80.96 72.83 76.68 advp 866 832 683 82.09 78.87 80.45 prt 106 86 69 80.23 65.09 71.88 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.55 64.10 67.62 Avg2. 23852 23796 22189 93.25 93.03 93.14 Current max chunk-based F1: 93.14 (iteration 26) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 27 Log-likelihood = -53487.471575 Norm (log-likelihood gradient vector) = 1825.096979 Norm (lambda vector) = 109.112640 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12362 12040 97.40 96.92 97.16 e-pp 4811 4900 4724 96.41 98.19 97.29 i-np 14376 14576 13983 95.93 97.27 96.59 i-vp 2646 2681 2533 94.48 95.73 95.10 e-vp 4658 4679 4470 95.53 95.96 95.75 e-sbar 535 501 462 92.22 86.36 89.19 o 6180 6085 5917 97.24 95.74 96.49 e-adjp 438 404 341 84.41 77.85 81.00 i-advp 89 75 51 68.00 57.30 62.20 e-advp 866 824 690 83.74 79.68 81.66 i-adjp 167 134 109 81.34 65.27 72.43 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 36 31 86.11 64.58 73.81 e-prt 106 88 69 78.41 65.09 71.13 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.18 71.30 73.66 Avg2. 47375 47375 45438 95.91 95.91 95.91 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12363 11573 93.61 93.17 93.39 pp 4811 4900 4715 96.22 98.00 97.11 vp 4658 4679 4357 93.12 93.54 93.33 sbar 535 501 450 89.82 84.11 86.87 adjp 438 404 324 80.20 73.97 76.96 advp 866 824 678 82.28 78.29 80.24 prt 106 88 69 78.41 65.09 71.13 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.37 64.17 67.58 Avg2. 23852 23764 22171 93.30 92.95 93.12 Current max chunk-based F1: 93.14 (iteration 26) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 28 Log-likelihood = -51580.913100 Norm (log-likelihood gradient vector) = 2328.457116 Norm (lambda vector) = 111.983754 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12450 12093 97.13 97.35 97.24 e-pp 4811 4893 4724 96.55 98.19 97.36 i-np 14376 14386 13904 96.65 96.72 96.68 i-vp 2646 2672 2528 94.61 95.54 95.07 e-vp 4658 4686 4481 95.63 96.20 95.91 e-sbar 535 502 463 92.23 86.54 89.30 o 6180 6196 5971 96.37 96.62 96.49 e-adjp 438 411 345 83.94 78.77 81.27 i-advp 89 71 50 70.42 56.18 62.50 e-advp 866 813 685 84.26 79.10 81.60 i-adjp 167 134 110 82.09 65.87 73.09 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 92 71 77.17 66.98 71.72 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.06 71.57 73.75 Avg2. 47375 47375 45475 95.99 95.99 95.99 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12452 11642 93.50 93.72 93.61 pp 4811 4893 4715 96.36 98.00 97.18 vp 4658 4686 4364 93.13 93.69 93.41 sbar 535 502 451 89.84 84.30 86.98 adjp 438 411 326 79.32 74.43 76.80 advp 866 813 672 82.66 77.60 80.05 prt 106 92 71 77.17 66.98 71.72 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.20 64.43 67.64 Avg2. 23852 23854 22246 93.26 93.27 93.26 Current max chunk-based F1: 93.26 (iteration 28) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 29 Log-likelihood = -49710.949496 Norm (log-likelihood gradient vector) = 2449.167950 Norm (lambda vector) = 116.881439 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12407 12079 97.36 97.24 97.30 e-pp 4811 4898 4724 96.45 98.19 97.31 i-np 14376 14475 13960 96.44 97.11 96.77 i-vp 2646 2673 2528 94.58 95.54 95.06 e-vp 4658 4694 4486 95.57 96.31 95.94 e-sbar 535 500 461 92.20 86.17 89.08 o 6180 6147 5951 96.81 96.29 96.55 e-adjp 438 410 345 84.15 78.77 81.37 i-advp 89 70 49 70.00 55.06 61.64 e-advp 866 807 683 84.63 78.87 81.65 i-adjp 167 130 109 83.85 65.27 73.40 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 42 33 78.57 68.75 73.33 e-prt 106 92 71 77.17 66.98 71.72 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.99 71.56 73.71 Avg2. 47375 47375 45497 96.04 96.04 96.04 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12409 11637 93.78 93.68 93.73 pp 4811 4898 4714 96.24 97.98 97.11 vp 4658 4694 4369 93.08 93.80 93.43 sbar 535 500 449 89.80 83.93 86.76 adjp 438 410 327 79.76 74.66 77.12 advp 866 807 669 82.90 77.25 79.98 prt 106 92 71 77.17 66.98 71.72 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.27 64.38 67.65 Avg2. 23852 23815 22241 93.39 93.25 93.32 Current max chunk-based F1: 93.32 (iteration 29) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 30 Log-likelihood = -48258.923384 Norm (log-likelihood gradient vector) = 1630.787099 Norm (lambda vector) = 120.038174 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12417 12080 97.29 97.25 97.27 e-pp 4811 4894 4717 96.38 98.05 97.21 i-np 14376 14446 13936 96.47 96.94 96.70 i-vp 2646 2682 2531 94.37 95.65 95.01 e-vp 4658 4703 4492 95.51 96.44 95.97 e-sbar 535 507 464 91.52 86.73 89.06 o 6180 6150 5947 96.70 96.23 96.46 e-adjp 438 402 342 85.07 78.08 81.43 i-advp 89 69 50 72.46 56.18 63.29 e-advp 866 820 690 84.15 79.68 81.85 i-adjp 167 127 108 85.04 64.67 73.47 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 44 33 75.00 68.75 71.74 e-prt 106 88 70 79.55 66.04 72.16 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 8 8 100.00 61.54 76.19 e-conjp 9 3 3 100.00 33.33 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.08 69.47 72.63 Avg2. 47375 47375 45474 95.99 95.99 95.99 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12419 11626 93.61 93.59 93.60 pp 4811 4894 4707 96.18 97.84 97.00 vp 4658 4703 4374 93.00 93.90 93.45 sbar 535 507 452 89.15 84.49 86.76 adjp 438 402 326 81.09 74.43 77.62 advp 866 820 679 82.80 78.41 80.55 prt 106 88 70 79.55 66.04 72.16 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 5 3 60.00 33.33 42.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 67.54 62.20 64.76 Avg2. 23852 23838 22237 93.28 93.23 93.26 Current max chunk-based F1: 93.32 (iteration 29) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 31 Log-likelihood = -46996.892992 Norm (log-likelihood gradient vector) = 1481.303965 Norm (lambda vector) = 122.923315 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12391 12062 97.34 97.10 97.22 e-pp 4811 4873 4711 96.68 97.92 97.29 i-np 14376 14508 13962 96.24 97.12 96.68 i-vp 2646 2676 2527 94.43 95.50 94.96 e-vp 4658 4706 4490 95.41 96.39 95.90 e-sbar 535 514 468 91.05 87.48 89.23 o 6180 6132 5937 96.82 96.07 96.44 e-adjp 438 396 341 86.11 77.85 81.77 i-advp 89 69 50 72.46 56.18 63.29 e-advp 866 833 691 82.95 79.79 81.34 i-adjp 167 124 105 84.68 62.87 72.16 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 44 33 75.00 68.75 71.74 e-prt 106 83 67 80.72 63.21 70.90 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 8 8 100.00 61.54 76.19 e-conjp 9 3 3 100.00 33.33 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.11 69.23 72.50 Avg2. 47375 47375 45458 95.95 95.95 95.95 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12396 11615 93.70 93.50 93.60 pp 4811 4873 4702 96.49 97.73 97.11 vp 4658 4706 4372 92.90 93.86 93.38 sbar 535 514 456 88.72 85.23 86.94 adjp 438 396 325 82.07 74.20 77.94 advp 866 833 679 81.51 78.41 79.93 prt 106 83 67 80.72 63.21 70.90 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 5 3 60.00 33.33 42.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 67.61 61.95 64.66 Avg2. 23852 23806 22219 93.33 93.15 93.24 Current max chunk-based F1: 93.32 (iteration 29) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 32 Log-likelihood = -45548.526633 Norm (log-likelihood gradient vector) = 1970.848512 Norm (lambda vector) = 126.607959 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12470 12099 97.02 97.40 97.21 e-pp 4811 4867 4709 96.75 97.88 97.31 i-np 14376 14343 13882 96.79 96.56 96.67 i-vp 2646 2682 2528 94.26 95.54 94.89 e-vp 4658 4710 4491 95.35 96.41 95.88 e-sbar 535 520 470 90.38 87.85 89.10 o 6180 6193 5964 96.30 96.50 96.40 e-adjp 438 384 333 86.72 76.03 81.02 i-advp 89 69 50 72.46 56.18 63.29 e-advp 866 869 704 81.01 81.29 81.15 i-adjp 167 120 102 85.00 61.08 71.08 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 44 33 75.00 68.75 71.74 e-prt 106 81 67 82.72 63.21 71.66 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 6 6 100.00 46.15 63.16 e-conjp 9 2 2 100.00 22.22 36.36 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.10 67.67 71.64 Avg2. 47375 47375 45443 95.92 95.92 95.92 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12475 11654 93.42 93.82 93.62 pp 4811 4867 4701 96.59 97.71 97.15 vp 4658 4710 4369 92.76 93.80 93.27 sbar 535 520 458 88.08 85.61 86.82 adjp 438 384 315 82.03 71.92 76.64 advp 866 869 691 79.52 79.79 79.65 prt 106 81 67 82.72 63.21 71.66 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 4 2 50.00 22.22 30.77 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 66.51 60.81 63.53 Avg2. 23852 23910 22257 93.09 93.31 93.20 Current max chunk-based F1: 93.32 (iteration 29) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 33 Log-likelihood = -43774.202602 Norm (log-likelihood gradient vector) = 2091.120681 Norm (lambda vector) = 132.195214 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12441 12086 97.15 97.30 97.22 e-pp 4811 4871 4711 96.72 97.92 97.31 i-np 14376 14424 13916 96.48 96.80 96.64 i-vp 2646 2661 2518 94.63 95.16 94.89 e-vp 4658 4705 4490 95.43 96.39 95.91 e-sbar 535 515 468 90.87 87.48 89.14 o 6180 6168 5951 96.48 96.29 96.39 e-adjp 438 378 329 87.04 75.11 80.64 i-advp 89 70 50 71.43 56.18 62.89 e-advp 866 878 709 80.75 81.87 81.31 i-adjp 167 118 100 84.75 59.88 70.18 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 44 32 72.73 66.67 69.57 e-prt 106 81 67 82.72 63.21 71.66 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 5 5 100.00 38.46 55.56 e-conjp 9 1 1 100.00 11.11 20.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.95 66.38 70.84 Avg2. 47375 47375 45436 95.91 95.91 95.91 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12446 11641 93.53 93.71 93.62 pp 4811 4871 4701 96.51 97.71 97.11 vp 4658 4705 4369 92.86 93.80 93.32 sbar 535 515 456 88.54 85.23 86.86 adjp 438 378 313 82.80 71.46 76.72 advp 866 878 697 79.38 80.48 79.93 prt 106 81 67 82.72 63.21 71.66 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 4 1 25.00 11.11 15.38 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 64.13 59.67 61.82 Avg2. 23852 23878 22245 93.16 93.26 93.21 Current max chunk-based F1: 93.32 (iteration 29) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 34 Log-likelihood = -42184.196771 Norm (log-likelihood gradient vector) = 1407.414992 Norm (lambda vector) = 135.614669 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12457 12101 97.14 97.42 97.28 e-pp 4811 4870 4708 96.67 97.86 97.26 i-np 14376 14374 13910 96.77 96.76 96.77 i-vp 2646 2657 2521 94.88 95.28 95.08 e-vp 4658 4698 4494 95.66 96.48 96.07 e-sbar 535 508 462 90.94 86.36 88.59 o 6180 6196 5965 96.27 96.52 96.40 e-adjp 438 399 341 85.46 77.85 81.48 i-advp 89 71 51 71.83 57.30 63.75 e-advp 866 871 710 81.52 81.99 81.75 i-adjp 167 129 107 82.95 64.07 72.30 i-sbar 4 14 3 21.43 75.00 33.33 i-pp 48 40 31 77.50 64.58 70.45 e-prt 106 85 69 81.18 65.09 72.25 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 5 5 100.00 38.46 55.56 e-conjp 9 1 1 100.00 11.11 20.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.12 66.79 71.15 Avg2. 47375 47375 45479 96.00 96.00 96.00 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12460 11669 93.65 93.94 93.79 pp 4811 4870 4701 96.53 97.71 97.12 vp 4658 4698 4373 93.08 93.88 93.48 sbar 535 508 451 88.78 84.30 86.48 adjp 438 399 322 80.70 73.52 76.94 advp 866 871 698 80.14 80.60 80.37 prt 106 85 69 81.18 65.09 72.25 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 4 1 25.00 11.11 15.38 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 63.91 60.02 61.90 Avg2. 23852 23895 22284 93.26 93.43 93.34 Current max chunk-based F1: 93.34 (iteration 34) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 35 Log-likelihood = -40523.859072 Norm (log-likelihood gradient vector) = 1511.130760 Norm (lambda vector) = 138.830732 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12398 12074 97.39 97.20 97.29 e-pp 4811 4870 4712 96.76 97.94 97.35 i-np 14376 14515 13969 96.24 97.17 96.70 i-vp 2646 2648 2519 95.13 95.20 95.16 e-vp 4658 4686 4488 95.77 96.35 96.06 e-sbar 535 517 467 90.33 87.29 88.78 o 6180 6131 5932 96.75 95.99 96.37 e-adjp 438 405 343 84.69 78.31 81.38 i-advp 89 74 51 68.92 57.30 62.58 e-advp 866 849 700 82.45 80.83 81.63 i-adjp 167 132 107 81.06 64.07 71.57 i-sbar 4 14 3 21.43 75.00 33.33 i-pp 48 42 31 73.81 64.58 68.89 e-prt 106 91 75 82.42 70.75 76.14 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 2 2 100.00 15.38 26.67 e-conjp 9 1 1 100.00 11.11 20.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.73 65.80 70.42 Avg2. 47375 47375 45474 95.99 95.99 95.99 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12401 11637 93.84 93.68 93.76 pp 4811 4870 4705 96.61 97.80 97.20 vp 4658 4686 4372 93.30 93.86 93.58 sbar 535 517 456 88.20 85.23 86.69 adjp 438 405 323 79.75 73.74 76.63 advp 866 849 689 81.15 79.56 80.35 prt 106 91 75 82.42 70.75 76.14 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 1 1 100.00 11.11 20.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.53 60.57 65.60 Avg2. 23852 23820 22258 93.44 93.32 93.38 Current max chunk-based F1: 93.38 (iteration 35) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 36 Log-likelihood = -38336.166163 Norm (log-likelihood gradient vector) = 1802.132294 Norm (lambda vector) = 144.130941 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12410 12094 97.45 97.36 97.41 e-pp 4811 4880 4721 96.74 98.13 97.43 i-np 14376 14483 13979 96.52 97.24 96.88 i-vp 2646 2649 2522 95.21 95.31 95.26 e-vp 4658 4691 4493 95.78 96.46 96.12 e-sbar 535 513 464 90.45 86.73 88.55 o 6180 6145 5949 96.81 96.26 96.54 e-adjp 438 411 343 83.45 78.31 80.80 i-advp 89 77 53 68.83 59.55 63.86 e-advp 866 824 692 83.98 79.91 81.89 i-adjp 167 134 107 79.85 64.07 71.10 i-sbar 4 14 3 21.43 75.00 33.33 i-pp 48 43 32 74.42 66.67 70.33 e-prt 106 98 79 80.61 74.53 77.45 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 2 2 100.00 15.38 26.67 e-conjp 9 1 1 100.00 11.11 20.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.64 66.22 70.62 Avg2. 47375 47375 45534 96.11 96.11 96.11 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12414 11667 93.98 93.92 93.95 pp 4811 4880 4714 96.60 97.98 97.29 vp 4658 4691 4377 93.31 93.97 93.64 sbar 535 513 453 88.30 84.67 86.45 adjp 438 411 323 78.59 73.74 76.09 advp 866 824 683 82.89 78.87 80.83 prt 106 98 79 80.61 74.53 77.45 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 1 1 100.00 11.11 20.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.43 60.88 65.73 Avg2. 23852 23832 22297 93.56 93.48 93.52 Current max chunk-based F1: 93.52 (iteration 36) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 37 Log-likelihood = -37129.718311 Norm (log-likelihood gradient vector) = 1229.451420 Norm (lambda vector) = 146.512316 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12403 12102 97.57 97.42 97.50 e-pp 4811 4870 4717 96.86 98.05 97.45 i-np 14376 14474 13990 96.66 97.31 96.98 i-vp 2646 2655 2525 95.10 95.43 95.27 e-vp 4658 4700 4494 95.62 96.48 96.05 e-sbar 535 524 469 89.50 87.66 88.57 o 6180 6142 5953 96.92 96.33 96.62 e-adjp 438 416 345 82.93 78.77 80.80 i-advp 89 81 54 66.67 60.67 63.53 e-advp 866 827 695 84.04 80.25 82.10 i-adjp 167 133 107 80.45 64.07 71.33 i-sbar 4 14 3 21.43 75.00 33.33 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 95 77 81.05 72.64 76.62 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 0 0 0.00 0.00 0.00 i-conjp 13 2 2 100.00 15.38 26.67 e-conjp 9 1 1 100.00 11.11 20.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.06 66.29 70.84 Avg2. 47375 47375 45566 96.18 96.18 96.18 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12406 11684 94.18 94.06 94.12 pp 4811 4870 4708 96.67 97.86 97.26 vp 4658 4700 4377 93.13 93.97 93.55 sbar 535 524 458 87.40 85.61 86.50 adjp 438 416 324 77.88 73.97 75.88 advp 866 827 685 82.83 79.10 80.92 prt 106 95 77 81.05 72.64 76.62 lst 5 0 0 0.00 0.00 0.00 intj 2 0 0 0.00 0.00 0.00 conjp 9 1 1 100.00 11.11 20.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 71.32 60.83 65.66 Avg2. 23852 23839 22314 93.60 93.55 93.58 Current max chunk-based F1: 93.58 (iteration 37) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 38 Log-likelihood = -35599.028117 Norm (log-likelihood gradient vector) = 1218.077208 Norm (lambda vector) = 150.188675 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12481 12133 97.21 97.67 97.44 e-pp 4811 4871 4718 96.86 98.07 97.46 i-np 14376 14339 13908 96.99 96.74 96.87 i-vp 2646 2667 2526 94.71 95.46 95.09 e-vp 4658 4687 4486 95.71 96.31 96.01 e-sbar 535 519 468 90.17 87.48 88.80 o 6180 6225 5986 96.16 96.86 96.51 e-adjp 438 412 346 83.98 79.00 81.41 i-advp 89 80 54 67.50 60.67 63.91 e-advp 866 816 689 84.44 79.56 81.93 i-adjp 167 123 104 84.55 62.28 71.72 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 98 78 79.59 73.58 76.47 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 2 2 100.00 15.38 26.67 e-conjp 9 1 1 100.00 11.11 20.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.78 68.99 74.84 Avg2. 47375 47375 45535 96.12 96.12 96.12 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12484 11701 93.73 94.20 93.96 pp 4811 4871 4709 96.67 97.88 97.27 vp 4658 4687 4366 93.15 93.73 93.44 sbar 535 519 456 87.86 85.23 86.53 adjp 438 412 326 79.13 74.43 76.71 advp 866 816 679 83.21 78.41 80.74 prt 106 98 78 79.59 73.58 76.47 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 1 1 100.00 11.11 20.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.33 65.86 72.78 Avg2. 23852 23889 22317 93.42 93.56 93.49 Current max chunk-based F1: 93.58 (iteration 37) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 39 Log-likelihood = -34045.893993 Norm (log-likelihood gradient vector) = 1907.206686 Norm (lambda vector) = 153.366727 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12452 12122 97.35 97.58 97.47 e-pp 4811 4871 4717 96.84 98.05 97.44 i-np 14376 14414 13962 96.86 97.12 96.99 i-vp 2646 2667 2530 94.86 95.62 95.24 e-vp 4658 4677 4486 95.92 96.31 96.11 e-sbar 535 521 470 90.21 87.85 89.02 o 6180 6180 5975 96.68 96.68 96.68 e-adjp 438 406 346 85.22 79.00 81.99 i-advp 89 78 54 69.23 60.67 64.67 e-advp 866 833 700 84.03 80.83 82.40 i-adjp 167 122 103 84.43 61.68 71.28 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 91 74 81.32 69.81 75.13 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 6 6 100.00 46.15 63.16 e-conjp 9 3 3 100.00 33.33 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.06 71.80 76.59 Avg2. 47375 47375 45584 96.22 96.22 96.22 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12455 11710 94.02 94.27 94.14 pp 4811 4871 4708 96.65 97.86 97.25 vp 4658 4677 4369 93.41 93.80 93.60 sbar 535 521 458 87.91 85.61 86.74 adjp 438 406 327 80.54 74.66 77.49 advp 866 833 689 82.71 79.56 81.11 prt 106 91 74 81.32 69.81 75.13 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 3 3 100.00 33.33 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.66 67.89 74.14 Avg2. 23852 23858 22339 93.63 93.66 93.64 Current max chunk-based F1: 93.64 (iteration 39) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 40 Log-likelihood = -32867.822014 Norm (log-likelihood gradient vector) = 1140.136302 Norm (lambda vector) = 155.590930 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12448 12118 97.35 97.55 97.45 e-pp 4811 4863 4711 96.87 97.92 97.40 i-np 14376 14412 13957 96.84 97.09 96.96 i-vp 2646 2659 2532 95.22 95.69 95.46 e-vp 4658 4682 4490 95.90 96.39 96.15 e-sbar 535 527 473 89.75 88.41 89.08 o 6180 6178 5974 96.70 96.67 96.68 e-adjp 438 402 344 85.57 78.54 81.90 i-advp 89 76 52 68.42 58.43 63.03 e-advp 866 860 716 83.26 82.68 82.97 i-adjp 167 117 103 88.03 61.68 72.54 i-sbar 4 14 3 21.43 75.00 33.33 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 89 72 80.90 67.92 73.85 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 6 6 100.00 46.15 63.16 e-conjp 9 3 3 100.00 33.33 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.25 71.67 76.60 Avg2. 47375 47375 45587 96.23 96.23 96.23 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12451 11708 94.03 94.25 94.14 pp 4811 4863 4703 96.71 97.76 97.23 vp 4658 4682 4380 93.55 94.03 93.79 sbar 535 527 462 87.67 86.36 87.01 adjp 438 402 326 81.09 74.43 77.62 advp 866 860 704 81.86 81.29 81.58 prt 106 89 72 80.90 67.92 73.85 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 3 3 100.00 33.33 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.58 67.94 74.14 Avg2. 23852 23878 22359 93.64 93.74 93.69 Current max chunk-based F1: 93.69 (iteration 40) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 41 Log-likelihood = -31813.308897 Norm (log-likelihood gradient vector) = 994.476447 Norm (lambda vector) = 157.880681 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12364 12060 97.54 97.09 97.31 e-pp 4811 4869 4716 96.86 98.03 97.44 i-np 14376 14579 14021 96.17 97.53 96.85 i-vp 2646 2663 2533 95.12 95.73 95.42 e-vp 4658 4674 4485 95.96 96.29 96.12 e-sbar 535 513 467 91.03 87.29 89.12 o 6180 6115 5936 97.07 96.05 96.56 e-adjp 438 393 339 86.26 77.40 81.59 i-advp 89 77 51 66.23 57.30 61.45 e-advp 866 861 713 82.81 82.33 82.57 i-adjp 167 114 101 88.60 60.48 71.89 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 89 73 82.02 68.87 74.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 6 6 100.00 46.15 63.16 e-conjp 9 3 3 100.00 33.33 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.24 71.53 76.51 Avg2. 47375 47375 45541 96.13 96.13 96.13 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12367 11631 94.05 93.63 93.84 pp 4811 4869 4709 96.71 97.88 97.29 vp 4658 4674 4371 93.52 93.84 93.68 sbar 535 513 455 88.69 85.05 86.83 adjp 438 393 321 81.68 73.29 77.26 advp 866 861 701 81.42 80.95 81.18 prt 106 89 73 82.02 68.87 74.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 3 3 100.00 33.33 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.81 67.68 74.08 Avg2. 23852 23770 22265 93.67 93.35 93.51 Current max chunk-based F1: 93.69 (iteration 40) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 42 Log-likelihood = -30857.841987 Norm (log-likelihood gradient vector) = 2064.593836 Norm (lambda vector) = 159.667806 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12417 12093 97.39 97.35 97.37 e-pp 4811 4866 4717 96.94 98.05 97.49 i-np 14376 14487 13990 96.57 97.31 96.94 i-vp 2646 2663 2532 95.08 95.69 95.39 e-vp 4658 4683 4488 95.84 96.35 96.09 e-sbar 535 513 470 91.62 87.85 89.69 o 6180 6147 5957 96.91 96.39 96.65 e-adjp 438 394 340 86.29 77.63 81.73 i-advp 89 76 51 67.11 57.30 61.82 e-advp 866 855 711 83.16 82.10 82.63 i-adjp 167 115 101 87.83 60.48 71.63 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 40 34 85.00 70.83 77.27 e-prt 106 91 75 82.42 70.75 76.14 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 8 8 100.00 61.54 76.19 e-conjp 9 4 4 100.00 44.44 61.54 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.34 73.28 77.55 Avg2. 47375 47375 45575 96.20 96.20 96.20 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12419 11674 94.00 93.98 93.99 pp 4811 4866 4711 96.81 97.92 97.36 vp 4658 4683 4374 93.40 93.90 93.65 sbar 535 513 458 89.28 85.61 87.40 adjp 438 394 321 81.47 73.29 77.16 advp 866 855 699 81.75 80.72 81.23 prt 106 91 75 82.42 70.75 76.14 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 4 4 100.00 44.44 61.54 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.91 69.06 74.94 Avg2. 23852 23826 22317 93.67 93.56 93.62 Current max chunk-based F1: 93.69 (iteration 40) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 43 Log-likelihood = -30111.896211 Norm (log-likelihood gradient vector) = 1127.681916 Norm (lambda vector) = 160.665103 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12449 12110 97.28 97.49 97.38 e-pp 4811 4861 4718 97.06 98.07 97.56 i-np 14376 14425 13958 96.76 97.09 96.93 i-vp 2646 2657 2537 95.48 95.88 95.68 e-vp 4658 4691 4493 95.78 96.46 96.12 e-sbar 535 514 472 91.83 88.22 89.99 o 6180 6174 5971 96.71 96.62 96.67 e-adjp 438 402 344 85.57 78.54 81.90 i-advp 89 74 51 68.92 57.30 62.58 e-advp 866 835 706 84.55 81.52 83.01 i-adjp 167 121 104 85.95 62.28 72.22 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 40 34 85.00 70.83 77.27 e-prt 106 101 81 80.20 76.42 78.26 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.28 75.23 78.60 Avg2. 47375 47375 45598 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12451 11686 93.86 94.08 93.97 pp 4811 4861 4712 96.93 97.94 97.44 vp 4658 4691 4384 93.46 94.12 93.79 sbar 535 514 460 89.49 85.98 87.70 adjp 438 402 324 80.60 73.97 77.14 advp 866 835 694 83.11 80.14 81.60 prt 106 101 81 80.20 76.42 78.26 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.76 70.82 75.90 Avg2. 23852 23861 22347 93.65 93.69 93.67 Current max chunk-based F1: 93.69 (iteration 40) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 44 Log-likelihood = -29117.988515 Norm (log-likelihood gradient vector) = 912.709597 Norm (lambda vector) = 162.720162 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12488 12127 97.11 97.63 97.37 e-pp 4811 4874 4723 96.90 98.17 97.53 i-np 14376 14352 13922 97.00 96.84 96.92 i-vp 2646 2652 2536 95.63 95.84 95.73 e-vp 4658 4693 4497 95.82 96.54 96.18 e-sbar 535 502 463 92.23 86.54 89.30 o 6180 6203 5986 96.50 96.86 96.68 e-adjp 438 409 346 84.60 79.00 81.70 i-advp 89 72 50 69.44 56.18 62.11 e-advp 866 824 702 85.19 81.06 83.08 i-adjp 167 129 107 82.95 64.07 72.30 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 41 35 85.37 72.92 78.65 e-prt 106 105 82 78.10 77.36 77.73 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 82.05 75.36 78.56 Avg2. 47375 47375 45595 96.24 96.24 96.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12490 11706 93.72 94.24 93.98 pp 4811 4874 4717 96.78 98.05 97.41 vp 4658 4693 4389 93.52 94.22 93.87 sbar 535 502 451 89.84 84.30 86.98 adjp 438 409 326 79.71 74.43 76.98 advp 866 824 690 83.74 79.68 81.66 prt 106 105 82 78.10 77.36 77.73 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.54 70.78 75.78 Avg2. 23852 23903 22367 93.57 93.77 93.67 Current max chunk-based F1: 93.69 (iteration 40) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 45 Log-likelihood = -28309.417269 Norm (log-likelihood gradient vector) = 1227.199959 Norm (lambda vector) = 164.619094 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12458 12117 97.26 97.54 97.40 e-pp 4811 4858 4716 97.08 98.03 97.55 i-np 14376 14407 13958 96.88 97.09 96.99 i-vp 2646 2647 2533 95.69 95.73 95.71 e-vp 4658 4703 4499 95.66 96.59 96.12 e-sbar 535 515 470 91.26 87.85 89.52 o 6180 6174 5975 96.78 96.68 96.73 e-adjp 438 406 346 85.22 79.00 81.99 i-advp 89 73 51 69.86 57.30 62.96 e-advp 866 821 702 85.51 81.06 83.22 i-adjp 167 135 111 82.22 66.47 73.51 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 42 35 83.33 72.92 77.78 e-prt 106 105 80 76.19 75.47 75.83 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 10 10 100.00 76.92 86.96 e-conjp 9 5 5 100.00 55.56 71.43 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.83 75.51 78.54 Avg2. 47375 47375 45612 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12460 11701 93.91 94.20 94.05 pp 4811 4858 4710 96.95 97.90 97.42 vp 4658 4703 4389 93.32 94.22 93.77 sbar 535 515 458 88.93 85.61 87.24 adjp 438 406 326 80.30 74.43 77.25 advp 866 821 691 84.17 79.79 81.92 prt 106 105 80 76.19 75.47 75.83 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 5 100.00 55.56 71.43 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.38 70.72 75.67 Avg2. 23852 23874 22361 93.66 93.75 93.71 Current max chunk-based F1: 93.71 (iteration 45) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 46 Log-likelihood = -27251.162598 Norm (log-likelihood gradient vector) = 1100.208621 Norm (lambda vector) = 167.424689 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12481 12125 97.15 97.61 97.38 e-pp 4811 4876 4723 96.86 98.17 97.51 i-np 14376 14357 13932 97.04 96.91 96.98 i-vp 2646 2654 2535 95.52 95.80 95.66 e-vp 4658 4695 4496 95.76 96.52 96.14 e-sbar 535 507 466 91.91 87.10 89.44 o 6180 6200 5981 96.47 96.78 96.62 e-adjp 438 399 343 85.96 78.31 81.96 i-advp 89 73 51 69.86 57.30 62.96 e-advp 866 825 703 85.21 81.18 83.15 i-adjp 167 135 110 81.48 65.87 72.85 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 42 35 83.33 72.92 77.78 e-prt 106 98 77 78.57 72.64 75.49 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 6 5 83.33 55.56 66.67 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.52 75.26 77.80 Avg2. 47375 47375 45596 96.24 96.24 96.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12483 11710 93.81 94.27 94.04 pp 4811 4876 4717 96.74 98.05 97.39 vp 4658 4695 4387 93.44 94.18 93.81 sbar 535 507 454 89.55 84.86 87.14 adjp 438 399 323 80.95 73.74 77.18 advp 866 825 691 83.76 79.79 81.73 prt 106 98 77 78.57 72.64 75.49 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.01 70.31 74.85 Avg2. 23852 23890 22365 93.62 93.77 93.69 Current max chunk-based F1: 93.71 (iteration 45) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 47 Log-likelihood = -26390.446169 Norm (log-likelihood gradient vector) = 1276.605477 Norm (lambda vector) = 170.226588 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12457 12117 97.27 97.54 97.41 e-pp 4811 4879 4722 96.78 98.15 97.46 i-np 14376 14400 13958 96.93 97.09 97.01 i-vp 2646 2654 2534 95.48 95.77 95.62 e-vp 4658 4692 4494 95.78 96.48 96.13 e-sbar 535 514 468 91.05 87.48 89.23 o 6180 6171 5971 96.76 96.62 96.69 e-adjp 438 396 339 85.61 77.40 81.29 i-advp 89 74 52 70.27 58.43 63.80 e-advp 866 837 708 84.59 81.76 83.15 i-adjp 167 134 109 81.34 65.27 72.43 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 43 35 81.40 72.92 76.92 e-prt 106 91 74 81.32 69.81 75.13 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 6 5 83.33 55.56 66.67 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.49 75.12 77.71 Avg2. 47375 47375 45600 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12458 11704 93.95 94.22 94.08 pp 4811 4879 4716 96.66 98.03 97.34 vp 4658 4692 4386 93.48 94.16 93.82 sbar 535 514 456 88.72 85.23 86.94 adjp 438 396 320 80.81 73.06 76.74 advp 866 837 696 83.15 80.37 81.74 prt 106 91 74 81.32 69.81 75.13 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.14 70.04 74.75 Avg2. 23852 23874 22358 93.65 93.74 93.69 Current max chunk-based F1: 93.71 (iteration 45) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 48 Log-likelihood = -25582.709665 Norm (log-likelihood gradient vector) = 901.456652 Norm (lambda vector) = 171.854268 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12444 12114 97.35 97.52 97.43 e-pp 4811 4876 4723 96.86 98.17 97.51 i-np 14376 14421 13970 96.87 97.18 97.02 i-vp 2646 2653 2532 95.44 95.69 95.57 e-vp 4658 4689 4497 95.91 96.54 96.22 e-sbar 535 527 477 90.51 89.16 89.83 o 6180 6166 5972 96.85 96.63 96.74 e-adjp 438 387 337 87.08 76.94 81.70 i-advp 89 76 54 71.05 60.67 65.45 e-advp 866 856 715 83.53 82.56 83.04 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 40 34 85.00 70.83 77.27 e-prt 106 84 71 84.52 66.98 74.74 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 10 90.91 76.92 83.33 e-conjp 9 6 5 83.33 55.56 66.67 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 81.14 74.96 77.93 Avg2. 47375 47375 45620 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12445 11700 94.01 94.19 94.10 pp 4811 4876 4716 96.72 98.03 97.37 vp 4658 4689 4390 93.62 94.25 93.93 sbar 535 527 465 88.24 86.92 87.57 adjp 438 387 320 82.69 73.06 77.58 advp 866 856 704 82.24 81.29 81.77 prt 106 84 71 84.52 66.98 74.74 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 5 83.33 55.56 66.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.54 70.03 74.92 Avg2. 23852 23871 22372 93.72 93.80 93.76 Current max chunk-based F1: 93.76 (iteration 48) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 49 Log-likelihood = -24585.686506 Norm (log-likelihood gradient vector) = 794.817757 Norm (lambda vector) = 174.106896 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12348 12048 97.57 96.99 97.28 e-pp 4811 4879 4727 96.88 98.25 97.56 i-np 14376 14643 14040 95.88 97.66 96.76 i-vp 2646 2656 2533 95.37 95.73 95.55 e-vp 4658 4680 4495 96.05 96.50 96.27 e-sbar 535 517 473 91.49 88.41 89.92 o 6180 6055 5904 97.51 95.53 96.51 e-adjp 438 390 337 86.41 76.94 81.40 i-advp 89 79 55 69.62 61.80 65.48 e-advp 866 848 711 83.84 82.10 82.96 i-adjp 167 121 105 86.78 62.87 72.92 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 40 34 85.00 70.83 77.27 e-prt 106 84 70 83.33 66.04 73.68 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 10 83.33 76.92 80.00 e-conjp 9 7 5 71.43 55.56 62.50 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 80.03 74.84 77.35 Avg2. 47375 47375 45551 96.15 96.15 96.15 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12348 11618 94.09 93.53 93.81 pp 4811 4879 4720 96.74 98.11 97.42 vp 4658 4680 4388 93.76 94.20 93.98 sbar 535 517 461 89.17 86.17 87.64 adjp 438 390 322 82.56 73.52 77.78 advp 866 848 698 82.31 80.60 81.45 prt 106 84 70 83.33 66.04 73.68 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 5 71.43 55.56 62.50 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 79.34 69.77 74.25 Avg2. 23852 23754 22283 93.81 93.42 93.61 Current max chunk-based F1: 93.76 (iteration 48) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 50 Log-likelihood = -24031.417257 Norm (log-likelihood gradient vector) = 2333.355608 Norm (lambda vector) = 178.439264 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12432 12111 97.42 97.50 97.46 e-pp 4811 4869 4722 96.98 98.15 97.56 i-np 14376 14463 13981 96.67 97.25 96.96 i-vp 2646 2660 2533 95.23 95.73 95.48 e-vp 4658 4687 4497 95.95 96.54 96.24 e-sbar 535 526 479 91.06 89.53 90.29 o 6180 6139 5957 97.04 96.39 96.71 e-adjp 438 393 339 86.26 77.40 81.59 i-advp 89 79 55 69.62 61.80 65.48 e-advp 866 851 711 83.55 82.10 82.82 i-adjp 167 125 107 85.60 64.07 73.29 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 83 70 84.34 66.04 74.07 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 8 6 75.00 46.15 57.14 e-conjp 9 5 3 60.00 33.33 42.86 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.85 71.99 75.26 Avg2. 47375 47375 45608 96.27 96.27 96.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12433 11697 94.08 94.16 94.12 pp 4811 4869 4715 96.84 98.00 97.42 vp 4658 4687 4389 93.64 94.22 93.93 sbar 535 526 467 88.78 87.29 88.03 adjp 438 393 324 82.44 73.97 77.98 advp 866 851 698 82.02 80.60 81.30 prt 106 83 70 84.34 66.04 74.07 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 3 60.00 33.33 42.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.21 67.76 72.61 Avg2. 23852 23848 22364 93.78 93.76 93.77 Current max chunk-based F1: 93.77 (iteration 50) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 51 Log-likelihood = -23397.953006 Norm (log-likelihood gradient vector) = 983.765085 Norm (lambda vector) = 178.768359 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12468 12131 97.30 97.66 97.48 e-pp 4811 4864 4722 97.08 98.15 97.61 i-np 14376 14401 13956 96.91 97.08 96.99 i-vp 2646 2658 2530 95.18 95.62 95.40 e-vp 4658 4689 4498 95.93 96.57 96.24 e-sbar 535 524 479 91.41 89.53 90.46 o 6180 6172 5974 96.79 96.67 96.73 e-adjp 438 396 341 86.11 77.85 81.77 i-advp 89 79 55 69.62 61.80 65.48 e-advp 866 841 707 84.07 81.64 82.84 i-adjp 167 126 107 84.92 64.07 73.04 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 89 76 85.39 71.70 77.95 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 8 6 75.00 46.15 57.14 e-conjp 9 5 3 60.00 33.33 42.86 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.91 72.31 75.46 Avg2. 47375 47375 45622 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12469 11722 94.01 94.36 94.19 pp 4811 4864 4715 96.94 98.00 97.47 vp 4658 4689 4388 93.58 94.20 93.89 sbar 535 524 467 89.12 87.29 88.20 adjp 438 396 325 82.07 74.20 77.94 advp 866 841 694 82.52 80.14 81.31 prt 106 89 76 85.39 71.70 77.95 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 3 60.00 33.33 42.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.36 68.32 73.00 Avg2. 23852 23878 22391 93.77 93.87 93.82 Current max chunk-based F1: 93.82 (iteration 51) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 52 Log-likelihood = -23061.919138 Norm (log-likelihood gradient vector) = 808.237597 Norm (lambda vector) = 178.790948 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12485 12134 97.19 97.68 97.43 e-pp 4811 4865 4724 97.10 98.19 97.64 i-np 14376 14353 13931 97.06 96.90 96.98 i-vp 2646 2657 2530 95.22 95.62 95.42 e-vp 4658 4696 4500 95.83 96.61 96.22 e-sbar 535 519 476 91.71 88.97 90.32 o 6180 6202 5986 96.52 96.86 96.69 e-adjp 438 398 342 85.93 78.08 81.82 i-advp 89 79 55 69.62 61.80 65.48 e-advp 866 834 705 84.53 81.41 82.94 i-adjp 167 127 107 84.25 64.07 72.79 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 92 76 82.61 71.70 76.77 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 8 6 75.00 46.15 57.14 e-conjp 9 5 3 60.00 33.33 42.86 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.73 72.29 75.37 Avg2. 47375 47375 45612 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12486 11722 93.88 94.36 94.12 pp 4811 4865 4717 96.96 98.05 97.50 vp 4658 4696 4392 93.53 94.29 93.91 sbar 535 519 464 89.40 86.73 88.05 adjp 438 398 325 81.66 74.20 77.75 advp 866 834 692 82.97 79.91 81.41 prt 106 92 76 82.61 71.70 76.77 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 3 60.00 33.33 42.86 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.10 68.26 72.85 Avg2. 23852 23896 22392 93.71 93.88 93.79 Current max chunk-based F1: 93.82 (iteration 51) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 53 Log-likelihood = -22722.259139 Norm (log-likelihood gradient vector) = 941.290977 Norm (lambda vector) = 179.587905 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12480 12128 97.18 97.63 97.41 e-pp 4811 4864 4724 97.12 98.19 97.65 i-np 14376 14360 13924 96.96 96.86 96.91 i-vp 2646 2663 2533 95.12 95.73 95.42 e-vp 4658 4694 4502 95.91 96.65 96.28 e-sbar 535 516 474 91.86 88.60 90.20 o 6180 6201 5981 96.45 96.78 96.62 e-adjp 438 400 345 86.25 78.77 82.34 i-advp 89 78 55 70.51 61.80 65.87 e-advp 866 817 697 85.31 80.48 82.83 i-adjp 167 129 109 84.50 65.27 73.65 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 103 82 79.61 77.36 78.47 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 9 6 66.67 46.15 54.55 e-conjp 9 6 3 50.00 33.33 40.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.67 72.63 75.07 Avg2. 47375 47375 45600 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12481 11711 93.83 94.28 94.05 pp 4811 4864 4717 96.98 98.05 97.51 vp 4658 4694 4390 93.52 94.25 93.88 sbar 535 516 462 89.53 86.36 87.92 adjp 438 400 328 82.00 74.89 78.28 advp 866 817 685 83.84 79.10 81.40 prt 106 103 82 79.61 77.36 78.47 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 3 50.00 33.33 40.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.93 68.76 72.62 Avg2. 23852 23882 22379 93.71 93.82 93.77 Current max chunk-based F1: 93.82 (iteration 51) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 54 Log-likelihood = -22163.099653 Norm (log-likelihood gradient vector) = 965.119042 Norm (lambda vector) = 181.247958 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12519 12146 97.02 97.78 97.40 e-pp 4811 4884 4735 96.95 98.42 97.68 i-np 14376 14252 13869 97.31 96.47 96.89 i-vp 2646 2674 2538 94.91 95.92 95.41 e-vp 4658 4705 4510 95.86 96.82 96.34 e-sbar 535 497 462 92.96 86.36 89.53 o 6180 6238 5995 96.10 97.01 96.55 e-adjp 438 401 343 85.54 78.31 81.76 i-advp 89 76 55 72.37 61.80 66.67 e-advp 866 823 700 85.05 80.83 82.89 i-adjp 167 130 109 83.85 65.27 73.40 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 41 33 80.49 68.75 74.16 e-prt 106 104 82 78.85 77.36 78.10 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 9 6 66.67 46.15 54.55 e-conjp 9 6 3 50.00 33.33 40.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.44 72.53 74.91 Avg2. 47375 47375 45590 96.23 96.23 96.23 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12520 11729 93.68 94.42 94.05 pp 4811 4884 4728 96.81 98.27 97.53 vp 4658 4705 4397 93.45 94.40 93.92 sbar 535 497 450 90.54 84.11 87.21 adjp 438 401 324 80.80 73.97 77.23 advp 866 823 688 83.60 79.45 81.47 prt 106 104 82 78.85 77.36 78.10 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 3 50.00 33.33 40.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.77 68.53 72.42 Avg2. 23852 23941 22402 93.57 93.92 93.75 Current max chunk-based F1: 93.82 (iteration 51) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 55 Log-likelihood = -21442.078144 Norm (log-likelihood gradient vector) = 1812.514777 Norm (lambda vector) = 185.350878 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12422 12114 97.52 97.52 97.52 e-pp 4811 4873 4733 97.13 98.38 97.75 i-np 14376 14435 13985 96.88 97.28 97.08 i-vp 2646 2666 2540 95.27 95.99 95.63 e-vp 4658 4700 4512 96.00 96.87 96.43 e-sbar 535 510 472 92.55 88.22 90.33 o 6180 6156 5964 96.88 96.50 96.69 e-adjp 438 403 347 86.10 79.22 82.52 i-advp 89 74 54 72.97 60.67 66.26 e-advp 866 827 705 85.25 81.41 83.28 i-adjp 167 128 108 84.38 64.67 73.22 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 45 34 75.56 70.83 73.12 e-prt 106 106 82 77.36 77.36 77.36 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 8 4 50.00 30.77 38.10 e-conjp 9 6 2 33.33 22.22 26.67 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.40 71.27 73.28 Avg2. 47375 47375 45660 96.38 96.38 96.38 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12423 11718 94.33 94.33 94.33 pp 4811 4873 4725 96.96 98.21 97.58 vp 4658 4700 4406 93.74 94.59 94.17 sbar 535 510 460 90.20 85.98 88.04 adjp 438 403 327 81.14 74.66 77.76 advp 866 827 694 83.92 80.14 81.98 prt 106 106 82 77.36 77.36 77.36 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 2 33.33 22.22 26.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.10 67.75 71.23 Avg2. 23852 23849 22415 93.99 93.98 93.98 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 56 Log-likelihood = -20734.068696 Norm (log-likelihood gradient vector) = 1180.744254 Norm (lambda vector) = 187.209886 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12424 12110 97.47 97.49 97.48 e-pp 4811 4879 4735 97.05 98.42 97.73 i-np 14376 14446 13985 96.81 97.28 97.04 i-vp 2646 2661 2538 95.38 95.92 95.65 e-vp 4658 4690 4506 96.08 96.74 96.41 e-sbar 535 510 472 92.55 88.22 90.33 o 6180 6151 5962 96.93 96.47 96.70 e-adjp 438 401 346 86.28 79.00 82.48 i-advp 89 74 54 72.97 60.67 66.26 e-advp 866 833 711 85.35 82.10 83.70 i-adjp 167 126 106 84.13 63.47 72.35 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 46 35 76.09 72.92 74.47 e-prt 106 104 80 76.92 75.47 76.19 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 8 4 50.00 30.77 38.10 e-conjp 9 6 2 33.33 22.22 26.67 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.41 71.23 73.26 Avg2. 47375 47375 45650 96.36 96.36 96.36 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12425 11715 94.29 94.31 94.30 pp 4811 4879 4727 96.88 98.25 97.56 vp 4658 4690 4401 93.84 94.48 94.16 sbar 535 510 460 90.20 85.98 88.04 adjp 438 401 326 81.30 74.43 77.71 advp 866 833 700 84.03 80.83 82.40 prt 106 104 80 76.92 75.47 76.19 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 2 33.33 22.22 26.67 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.08 67.60 71.14 Avg2. 23852 23849 22412 93.97 93.96 93.97 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 57 Log-likelihood = -20264.669790 Norm (log-likelihood gradient vector) = 644.333559 Norm (lambda vector) = 187.838757 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12423 12108 97.46 97.47 97.47 e-pp 4811 4883 4732 96.91 98.36 97.63 i-np 14376 14459 13987 96.74 97.29 97.01 i-vp 2646 2653 2532 95.44 95.69 95.57 e-vp 4658 4687 4503 96.07 96.67 96.37 e-sbar 535 514 471 91.63 88.04 89.80 o 6180 6141 5957 97.00 96.39 96.70 e-adjp 438 399 345 86.47 78.77 82.44 i-advp 89 74 54 72.97 60.67 66.26 e-advp 866 838 712 84.96 82.22 83.57 i-adjp 167 127 105 82.68 62.87 71.43 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 43 35 81.40 72.92 76.92 e-prt 106 101 79 78.22 74.53 76.33 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 10 6 60.00 46.15 52.17 e-conjp 9 7 3 42.86 33.33 37.50 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.71 72.58 74.59 Avg2. 47375 47375 45633 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12424 11712 94.27 94.28 94.28 pp 4811 4883 4725 96.76 98.21 97.48 vp 4658 4687 4398 93.83 94.42 94.13 sbar 535 514 459 89.30 85.79 87.51 adjp 438 399 325 81.45 74.20 77.66 advp 866 838 701 83.65 80.95 82.28 prt 106 101 79 78.22 74.53 76.33 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 3 42.86 33.33 37.50 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.03 68.57 72.11 Avg2. 23852 23854 22403 93.92 93.93 93.92 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 58 Log-likelihood = -19865.102940 Norm (log-likelihood gradient vector) = 648.932762 Norm (lambda vector) = 188.858605 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12459 12122 97.30 97.58 97.44 e-pp 4811 4900 4735 96.63 98.42 97.52 i-np 14376 14385 13948 96.96 97.02 96.99 i-vp 2646 2655 2534 95.44 95.77 95.60 e-vp 4658 4692 4505 96.01 96.72 96.36 e-sbar 535 516 473 91.67 88.41 90.01 o 6180 6164 5968 96.82 96.57 96.69 e-adjp 438 395 344 87.09 78.54 82.59 i-advp 89 76 55 72.37 61.80 66.67 e-advp 866 836 709 84.81 81.87 83.31 i-adjp 167 122 104 85.25 62.28 71.97 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 43 35 81.40 72.92 76.92 e-prt 106 96 75 78.12 70.75 74.26 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 8 66.67 61.54 64.00 e-conjp 9 8 4 50.00 44.44 47.06 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.59 73.87 75.68 Avg2. 47375 47375 45623 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12460 11714 94.01 94.30 94.16 pp 4811 4900 4728 96.49 98.27 97.37 vp 4658 4692 4400 93.78 94.46 94.12 sbar 535 516 461 89.34 86.17 87.73 adjp 438 395 324 82.03 73.97 77.79 advp 866 836 698 83.49 80.60 82.02 prt 106 96 75 78.12 70.75 74.26 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 4 50.00 44.44 47.06 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.73 69.30 72.82 Avg2. 23852 23904 22405 93.73 93.93 93.83 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 59 Log-likelihood = -19172.438055 Norm (log-likelihood gradient vector) = 1147.622394 Norm (lambda vector) = 192.080838 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12416 12102 97.47 97.42 97.45 e-pp 4811 4892 4730 96.69 98.32 97.50 i-np 14376 14458 13984 96.72 97.27 97.00 i-vp 2646 2649 2529 95.47 95.58 95.52 e-vp 4658 4696 4509 96.02 96.80 96.41 e-sbar 535 519 473 91.14 88.41 89.75 o 6180 6140 5953 96.95 96.33 96.64 e-adjp 438 397 345 86.90 78.77 82.63 i-advp 89 76 55 72.37 61.80 66.67 e-advp 866 838 710 84.73 81.99 83.33 i-adjp 167 124 106 85.48 63.47 72.85 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 42 34 80.95 70.83 75.56 e-prt 106 92 74 80.43 69.81 74.75 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 8 66.67 61.54 64.00 e-conjp 9 8 4 50.00 44.44 47.06 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.67 73.77 75.67 Avg2. 47375 47375 45620 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12417 11699 94.22 94.18 94.20 pp 4811 4892 4723 96.55 98.17 97.35 vp 4658 4696 4398 93.65 94.42 94.03 sbar 535 519 461 88.82 86.17 87.48 adjp 438 397 325 81.86 74.20 77.84 advp 866 838 699 83.41 80.72 82.04 prt 106 92 74 80.43 69.81 74.75 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 4 50.00 44.44 47.06 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.90 69.21 72.85 Avg2. 23852 23860 22384 93.81 93.85 93.83 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 60 Log-likelihood = -18621.324700 Norm (log-likelihood gradient vector) = 872.200437 Norm (lambda vector) = 194.473181 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12436 12117 97.43 97.54 97.49 e-pp 4811 4882 4726 96.80 98.23 97.51 i-np 14376 14419 13975 96.92 97.21 97.07 i-vp 2646 2653 2530 95.36 95.62 95.49 e-vp 4658 4695 4511 96.08 96.84 96.46 e-sbar 535 521 474 90.98 88.60 89.77 o 6180 6166 5969 96.81 96.59 96.70 e-adjp 438 401 347 86.53 79.22 82.72 i-advp 89 77 55 71.43 61.80 66.27 e-advp 866 829 705 85.04 81.41 83.19 i-adjp 167 125 106 84.80 63.47 72.60 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 42 34 80.95 70.83 75.56 e-prt 106 93 74 79.57 69.81 74.37 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 8 66.67 61.54 64.00 e-conjp 9 8 4 50.00 44.44 47.06 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.52 73.79 75.61 Avg2. 47375 47375 45639 96.34 96.34 96.34 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12437 11722 94.25 94.36 94.31 pp 4811 4882 4719 96.66 98.09 97.37 vp 4658 4695 4399 93.70 94.44 94.07 sbar 535 521 462 88.68 86.36 87.50 adjp 438 401 327 81.55 74.66 77.95 advp 866 829 694 83.72 80.14 81.89 prt 106 93 74 79.57 69.81 74.37 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 4 50.00 44.44 47.06 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.81 69.23 72.82 Avg2. 23852 23867 22402 93.86 93.92 93.89 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 61 Log-likelihood = -18294.155605 Norm (log-likelihood gradient vector) = 556.645540 Norm (lambda vector) = 195.673629 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12442 12122 97.43 97.58 97.51 e-pp 4811 4873 4723 96.92 98.17 97.54 i-np 14376 14404 13967 96.97 97.15 97.06 i-vp 2646 2664 2541 95.38 96.03 95.71 e-vp 4658 4698 4514 96.08 96.91 96.49 e-sbar 535 519 472 90.94 88.22 89.56 o 6180 6174 5973 96.74 96.65 96.70 e-adjp 438 404 349 86.39 79.68 82.90 i-advp 89 78 56 71.79 62.92 67.07 e-advp 866 811 700 86.31 80.83 83.48 i-adjp 167 128 108 84.38 64.67 73.22 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 102 82 80.39 77.36 78.85 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 8 61.54 61.54 61.54 e-conjp 9 9 4 44.44 44.44 44.44 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.98 74.10 75.52 Avg2. 47375 47375 45655 96.37 96.37 96.37 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12443 11726 94.24 94.40 94.32 pp 4811 4873 4715 96.76 98.00 97.38 vp 4658 4698 4410 93.87 94.68 94.27 sbar 535 519 460 88.63 85.98 87.29 adjp 438 404 329 81.44 75.11 78.15 advp 866 811 691 85.20 79.79 82.41 prt 106 102 82 80.39 77.36 78.85 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 4 44.44 44.44 44.44 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.50 69.98 73.09 Avg2. 23852 23860 22418 93.96 93.99 93.97 Current max chunk-based F1: 93.98 (iteration 55) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 62 Log-likelihood = -17957.077771 Norm (log-likelihood gradient vector) = 573.085978 Norm (lambda vector) = 197.619549 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12437 12118 97.44 97.55 97.49 e-pp 4811 4857 4721 97.20 98.13 97.66 i-np 14376 14404 13965 96.95 97.14 97.05 i-vp 2646 2663 2538 95.31 95.92 95.61 e-vp 4658 4696 4515 96.15 96.93 96.54 e-sbar 535 520 473 90.96 88.41 89.67 o 6180 6172 5971 96.74 96.62 96.68 e-adjp 438 405 349 86.17 79.68 82.80 i-advp 89 80 56 70.00 62.92 66.27 e-advp 866 823 708 86.03 81.76 83.84 i-adjp 167 134 111 82.84 66.47 73.75 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 105 83 79.05 78.30 78.67 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 8 57.14 61.54 59.26 e-conjp 9 10 4 40.00 44.44 42.11 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.33 74.30 75.31 Avg2. 47375 47375 45656 96.37 96.37 96.37 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12438 11722 94.24 94.36 94.30 pp 4811 4857 4713 97.04 97.96 97.50 vp 4658 4696 4407 93.85 94.61 94.23 sbar 535 520 461 88.65 86.17 87.39 adjp 438 405 330 81.48 75.34 78.29 advp 866 823 698 84.81 80.60 82.65 prt 106 105 83 79.05 78.30 78.67 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 4 40.00 44.44 42.11 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.91 70.18 72.93 Avg2. 23852 23855 22419 93.98 93.99 93.99 Current max chunk-based F1: 93.99 (iteration 62) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 63 Log-likelihood = -17513.435276 Norm (log-likelihood gradient vector) = 564.693971 Norm (lambda vector) = 200.513450 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12429 12118 97.50 97.55 97.53 e-pp 4811 4869 4731 97.17 98.34 97.75 i-np 14376 14440 13988 96.87 97.30 97.08 i-vp 2646 2663 2538 95.31 95.92 95.61 e-vp 4658 4683 4510 96.31 96.82 96.56 e-sbar 535 500 465 93.00 86.92 89.86 o 6180 6173 5973 96.76 96.65 96.71 e-adjp 438 396 343 86.62 78.31 82.25 i-advp 89 81 57 70.37 64.04 67.06 e-advp 866 823 706 85.78 81.52 83.60 i-adjp 167 135 112 82.96 67.07 74.17 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 37 31 83.78 64.58 72.94 e-prt 106 106 86 81.13 81.13 81.13 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 8 57.14 61.54 59.26 e-conjp 9 10 4 40.00 44.44 42.11 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.71 74.29 75.48 Avg2. 47375 47375 45674 96.41 96.41 96.41 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12430 11730 94.37 94.43 94.40 pp 4811 4869 4723 97.00 98.17 97.58 vp 4658 4683 4399 93.94 94.44 94.19 sbar 535 500 453 90.60 84.67 87.54 adjp 438 396 324 81.82 73.97 77.70 advp 866 823 696 84.57 80.37 82.42 prt 106 106 86 81.13 81.13 81.13 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 4 40.00 44.44 42.11 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.34 70.16 73.12 Avg2. 23852 23818 22416 94.11 93.98 94.05 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 64 Log-likelihood = -16868.394925 Norm (log-likelihood gradient vector) = 1082.665796 Norm (lambda vector) = 206.440400 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12441 12122 97.44 97.58 97.51 e-pp 4811 4869 4730 97.15 98.32 97.73 i-np 14376 14405 13966 96.95 97.15 97.05 i-vp 2646 2666 2542 95.35 96.07 95.71 e-vp 4658 4688 4514 96.29 96.91 96.60 e-sbar 535 512 471 91.99 88.04 89.97 o 6180 6169 5971 96.79 96.62 96.70 e-adjp 438 399 346 86.72 79.00 82.68 i-advp 89 81 57 70.37 64.04 67.06 e-advp 866 830 709 85.42 81.87 83.61 i-adjp 167 134 111 82.84 66.47 73.75 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 103 84 81.55 79.25 80.38 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 8 57.14 61.54 59.26 e-conjp 9 10 4 40.00 44.44 42.11 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.68 74.39 75.51 Avg2. 47375 47375 45671 96.40 96.40 96.40 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12442 11731 94.29 94.44 94.36 pp 4811 4869 4722 96.98 98.15 97.56 vp 4658 4688 4403 93.92 94.53 94.22 sbar 535 512 459 89.65 85.79 87.68 adjp 438 399 325 81.45 74.20 77.66 advp 866 830 699 84.22 80.72 82.43 prt 106 103 84 81.55 79.25 80.38 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 4 40.00 44.44 42.11 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.21 70.15 73.05 Avg2. 23852 23854 22428 94.02 94.03 94.03 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 65 Log-likelihood = -16346.693082 Norm (log-likelihood gradient vector) = 563.596360 Norm (lambda vector) = 208.209524 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12426 12115 97.50 97.53 97.51 e-pp 4811 4867 4725 97.08 98.21 97.64 i-np 14376 14432 13982 96.88 97.26 97.07 i-vp 2646 2671 2541 95.13 96.03 95.58 e-vp 4658 4691 4514 96.23 96.91 96.57 e-sbar 535 519 473 91.14 88.41 89.75 o 6180 6150 5963 96.96 96.49 96.72 e-adjp 438 399 348 87.22 79.45 83.15 i-advp 89 80 55 68.75 61.80 65.09 e-advp 866 833 707 84.87 81.64 83.23 i-adjp 167 130 111 85.38 66.47 74.75 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 36 32 88.89 66.67 76.19 e-prt 106 98 81 82.65 76.42 79.41 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 16 10 62.50 76.92 68.97 e-conjp 9 11 5 45.45 55.56 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.59 75.60 76.58 Avg2. 47375 47375 45666 96.39 96.39 96.39 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12427 11723 94.33 94.37 94.35 pp 4811 4867 4717 96.92 98.05 97.48 vp 4658 4691 4399 93.78 94.44 94.11 sbar 535 519 461 88.82 86.17 87.48 adjp 438 399 328 82.21 74.89 78.38 advp 866 833 696 83.55 80.37 81.93 prt 106 98 81 82.65 76.42 79.41 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 11 5 45.45 55.56 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.77 71.03 73.79 Avg2. 23852 23846 22411 93.98 93.96 93.97 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 66 Log-likelihood = -16000.198127 Norm (log-likelihood gradient vector) = 520.747887 Norm (lambda vector) = 208.550736 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12442 12120 97.41 97.57 97.49 e-pp 4811 4874 4728 97.00 98.27 97.64 i-np 14376 14383 13958 97.05 97.09 97.07 i-vp 2646 2668 2536 95.05 95.84 95.45 e-vp 4658 4696 4514 96.12 96.91 96.51 e-sbar 535 521 474 90.98 88.60 89.77 o 6180 6170 5969 96.74 96.59 96.66 e-adjp 438 396 346 87.37 79.00 82.97 i-advp 89 80 55 68.75 61.80 65.09 e-advp 866 845 711 84.14 82.10 83.11 i-adjp 167 128 110 85.94 65.87 74.58 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 36 32 88.89 66.67 76.19 e-prt 106 93 78 83.87 73.58 78.39 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 16 10 62.50 76.92 68.97 e-conjp 9 11 5 45.45 55.56 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.63 75.41 76.50 Avg2. 47375 47375 45650 96.36 96.36 96.36 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12443 11730 94.27 94.43 94.35 pp 4811 4874 4720 96.84 98.11 97.47 vp 4658 4696 4395 93.59 94.35 93.97 sbar 535 521 462 88.68 86.36 87.50 adjp 438 396 326 82.32 74.43 78.18 advp 866 845 700 82.84 80.83 81.82 prt 106 93 78 83.87 73.58 78.39 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 11 5 45.45 55.56 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.79 70.76 73.65 Avg2. 23852 23880 22417 93.87 93.98 93.93 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 67 Log-likelihood = -15534.833135 Norm (log-likelihood gradient vector) = 729.146848 Norm (lambda vector) = 210.087935 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12388 12090 97.59 97.33 97.46 e-pp 4811 4861 4722 97.14 98.15 97.64 i-np 14376 14498 14016 96.68 97.50 97.08 i-vp 2646 2668 2533 94.94 95.73 95.33 e-vp 4658 4698 4509 95.98 96.80 96.39 e-sbar 535 529 478 90.36 89.35 89.85 o 6180 6124 5950 97.16 96.28 96.72 e-adjp 438 400 345 86.25 78.77 82.34 i-advp 89 80 55 68.75 61.80 65.09 e-advp 866 836 705 84.33 81.41 82.84 i-adjp 167 126 109 86.51 65.27 74.40 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 36 32 88.89 66.67 76.19 e-prt 106 93 78 83.87 73.58 78.39 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 8 61.54 61.54 61.54 e-conjp 9 9 4 44.44 44.44 44.44 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.47 73.87 75.62 Avg2. 47375 47375 45638 96.33 96.33 96.33 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12388 11701 94.45 94.20 94.32 pp 4811 4861 4714 96.98 97.98 97.48 vp 4658 4698 4390 93.44 94.25 93.84 sbar 535 529 466 88.09 87.10 87.59 adjp 438 400 326 81.50 74.43 77.80 advp 866 836 694 83.01 80.14 81.55 prt 106 93 78 83.87 73.58 78.39 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 4 44.44 44.44 44.44 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.58 69.61 72.93 Avg2. 23852 23815 22374 93.95 93.80 93.88 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 68 Log-likelihood = -15106.436367 Norm (log-likelihood gradient vector) = 1019.472179 Norm (lambda vector) = 212.844020 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12418 12104 97.47 97.44 97.46 e-pp 4811 4866 4727 97.14 98.25 97.70 i-np 14376 14442 13981 96.81 97.25 97.03 i-vp 2646 2667 2533 94.98 95.73 95.35 e-vp 4658 4694 4507 96.02 96.76 96.39 e-sbar 535 525 478 91.05 89.35 90.19 o 6180 6153 5958 96.83 96.41 96.62 e-adjp 438 401 345 86.03 78.77 82.24 i-advp 89 80 55 68.75 61.80 65.09 e-advp 866 833 702 84.27 81.06 82.64 i-adjp 167 126 109 86.51 65.27 74.40 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 36 32 88.89 66.67 76.19 e-prt 106 94 78 82.98 73.58 78.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 8 57.14 61.54 59.26 e-conjp 9 10 4 40.00 44.44 42.11 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.94 73.85 75.36 Avg2. 47375 47375 45625 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12418 11710 94.30 94.27 94.28 pp 4811 4866 4719 96.98 98.09 97.53 vp 4658 4694 4388 93.48 94.20 93.84 sbar 535 525 466 88.76 87.10 87.92 adjp 438 401 326 81.30 74.43 77.71 advp 866 833 691 82.95 79.79 81.34 prt 106 94 78 82.98 73.58 78.00 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 4 40.00 44.44 42.11 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.07 69.59 72.69 Avg2. 23852 23842 22383 93.88 93.84 93.86 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 69 Log-likelihood = -14900.874318 Norm (log-likelihood gradient vector) = 553.076993 Norm (lambda vector) = 213.548797 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12446 12120 97.38 97.57 97.47 e-pp 4811 4864 4728 97.20 98.27 97.74 i-np 14376 14395 13963 97.00 97.13 97.06 i-vp 2646 2666 2532 94.97 95.69 95.33 e-vp 4658 4695 4509 96.04 96.80 96.42 e-sbar 535 522 476 91.19 88.97 90.07 o 6180 6180 5970 96.60 96.60 96.60 e-adjp 438 402 345 85.82 78.77 82.14 i-advp 89 80 55 68.75 61.80 65.09 e-advp 866 826 698 84.50 80.60 82.51 i-adjp 167 130 111 85.38 66.47 74.75 i-sbar 4 15 3 20.00 75.00 31.58 i-pp 48 36 31 86.11 64.58 73.81 e-prt 106 97 79 81.44 74.53 77.83 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 8 66.67 61.54 64.00 e-conjp 9 8 4 50.00 44.44 47.06 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.73 73.82 75.72 Avg2. 47375 47375 45633 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12446 11731 94.26 94.44 94.35 pp 4811 4864 4720 97.04 98.11 97.57 vp 4658 4695 4390 93.50 94.25 93.87 sbar 535 522 464 88.89 86.73 87.80 adjp 438 402 327 81.34 74.66 77.86 advp 866 826 687 83.17 79.33 81.21 prt 106 97 79 81.44 74.53 77.83 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 4 50.00 44.44 47.06 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.96 69.65 73.12 Avg2. 23852 23861 22403 93.89 93.93 93.91 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 70 Log-likelihood = -14635.106296 Norm (log-likelihood gradient vector) = 439.975852 Norm (lambda vector) = 215.271471 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12465 12131 97.32 97.66 97.49 e-pp 4811 4865 4728 97.18 98.27 97.73 i-np 14376 14374 13953 97.07 97.06 97.06 i-vp 2646 2664 2531 95.01 95.65 95.33 e-vp 4658 4694 4509 96.06 96.80 96.43 e-sbar 535 520 475 91.35 88.79 90.05 o 6180 6187 5976 96.59 96.70 96.64 e-adjp 438 403 346 85.86 79.00 82.28 i-advp 89 80 55 68.75 61.80 65.09 e-advp 866 822 694 84.43 80.14 82.23 i-adjp 167 128 111 86.72 66.47 75.25 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 31 83.78 64.58 72.94 e-prt 106 102 80 78.43 75.47 76.92 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 10 6 60.00 46.15 52.17 e-conjp 9 7 3 42.86 33.33 37.50 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.68 72.38 74.47 Avg2. 47375 47375 45633 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12465 11740 94.18 94.51 94.35 pp 4811 4865 4720 97.02 98.11 97.56 vp 4658 4694 4390 93.52 94.25 93.88 sbar 535 520 462 88.85 86.36 87.58 adjp 438 403 329 81.64 75.11 78.24 advp 866 822 683 83.09 78.87 80.92 prt 106 102 80 78.43 75.47 76.92 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 3 42.86 33.33 37.50 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.96 68.60 72.09 Avg2. 23852 23879 22408 93.84 93.95 93.89 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 71 Log-likelihood = -14409.674346 Norm (log-likelihood gradient vector) = 530.194612 Norm (lambda vector) = 216.875560 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12484 12141 97.25 97.74 97.49 e-pp 4811 4862 4726 97.20 98.23 97.72 i-np 14376 14339 13937 97.20 96.95 97.07 i-vp 2646 2653 2528 95.29 95.54 95.41 e-vp 4658 4692 4508 96.08 96.78 96.43 e-sbar 535 523 475 90.82 88.79 89.79 o 6180 6205 5982 96.41 96.80 96.60 e-adjp 438 409 350 85.57 79.91 82.64 i-advp 89 79 55 69.62 61.80 65.48 e-advp 866 826 699 84.62 80.72 82.62 i-adjp 167 125 109 87.20 65.27 74.66 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 31 83.78 64.58 72.94 e-prt 106 107 82 76.64 77.36 77.00 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 10 6 60.00 46.15 52.17 e-conjp 9 7 3 42.86 33.33 37.50 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.63 72.50 74.50 Avg2. 47375 47375 45636 96.33 96.33 96.33 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12484 11750 94.12 94.59 94.35 pp 4811 4862 4718 97.04 98.07 97.55 vp 4658 4692 4392 93.61 94.29 93.95 sbar 535 523 462 88.34 86.36 87.33 adjp 438 409 331 80.93 75.57 78.16 advp 866 826 689 83.41 79.56 81.44 prt 106 107 82 76.64 77.36 77.00 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 3 42.86 33.33 37.50 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.69 68.91 72.14 Avg2. 23852 23911 22428 93.80 94.03 93.91 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 72 Log-likelihood = -13925.791103 Norm (log-likelihood gradient vector) = 734.865706 Norm (lambda vector) = 220.741859 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12428 12114 97.47 97.52 97.50 e-pp 4811 4887 4735 96.89 98.42 97.65 i-np 14376 14470 13996 96.72 97.36 97.04 i-vp 2646 2641 2523 95.53 95.35 95.44 e-vp 4658 4680 4497 96.09 96.54 96.32 e-sbar 535 493 459 93.10 85.79 89.30 o 6180 6158 5959 96.77 96.42 96.60 e-adjp 438 401 347 86.53 79.22 82.72 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 833 704 84.51 81.29 82.87 i-adjp 167 120 105 87.50 62.87 73.17 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 35 31 88.57 64.58 74.70 e-prt 106 109 82 75.23 77.36 76.28 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.97 75.17 77.02 Avg2. 47375 47375 45627 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12428 11713 94.25 94.29 94.27 pp 4811 4887 4727 96.73 98.25 97.48 vp 4658 4680 4386 93.72 94.16 93.94 sbar 535 493 446 90.47 83.36 86.77 adjp 438 401 327 81.55 74.66 77.95 advp 866 833 693 83.19 80.02 81.58 prt 106 109 82 75.23 77.36 76.28 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.76 70.77 74.10 Avg2. 23852 23840 22380 93.88 93.83 93.85 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 73 Log-likelihood = -13577.054396 Norm (log-likelihood gradient vector) = 1078.702176 Norm (lambda vector) = 226.425621 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12447 12126 97.42 97.62 97.52 e-pp 4811 4872 4732 97.13 98.36 97.74 i-np 14376 14403 13969 96.99 97.17 97.08 i-vp 2646 2645 2526 95.50 95.46 95.48 e-vp 4658 4692 4508 96.08 96.78 96.43 e-sbar 535 507 469 92.50 87.66 90.02 o 6180 6186 5974 96.57 96.67 96.62 e-adjp 438 403 348 86.35 79.45 82.76 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 835 706 84.55 81.52 83.01 i-adjp 167 121 106 87.60 63.47 73.61 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 35 31 88.57 64.58 74.70 e-prt 106 109 82 75.23 77.36 76.28 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.94 75.36 77.11 Avg2. 47375 47375 45652 96.36 96.36 96.36 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12447 11730 94.24 94.43 94.33 pp 4811 4872 4724 96.96 98.19 97.57 vp 4658 4692 4394 93.65 94.33 93.99 sbar 535 507 456 89.94 85.23 87.52 adjp 438 403 329 81.64 75.11 78.24 advp 866 835 695 83.23 80.25 81.72 prt 106 109 82 75.23 77.36 76.28 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.74 71.05 74.24 Avg2. 23852 23874 22416 93.89 93.98 93.94 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 74 Log-likelihood = -13210.962200 Norm (log-likelihood gradient vector) = 502.365701 Norm (lambda vector) = 226.243629 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12430 12120 97.51 97.57 97.54 e-pp 4811 4854 4721 97.26 98.13 97.69 i-np 14376 14420 13983 96.97 97.27 97.12 i-vp 2646 2644 2525 95.50 95.43 95.46 e-vp 4658 4698 4509 95.98 96.80 96.39 e-sbar 535 523 473 90.44 88.41 89.41 o 6180 6182 5973 96.62 96.65 96.63 e-adjp 438 403 348 86.35 79.45 82.76 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 838 706 84.25 81.52 82.86 i-adjp 167 122 107 87.70 64.07 74.05 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 35 31 88.57 64.58 74.70 e-prt 106 106 81 76.42 76.42 76.42 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.89 75.37 77.09 Avg2. 47375 47375 45652 96.36 96.36 96.36 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12430 11729 94.36 94.42 94.39 pp 4811 4854 4713 97.10 97.96 97.53 vp 4658 4698 4395 93.55 94.35 93.95 sbar 535 523 460 87.95 85.98 86.96 adjp 438 403 330 81.89 75.34 78.48 advp 866 838 695 82.94 80.25 81.57 prt 106 106 81 76.42 76.42 76.42 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.67 71.03 74.20 Avg2. 23852 23861 22409 93.91 93.95 93.93 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 75 Log-likelihood = -12992.880574 Norm (log-likelihood gradient vector) = 402.476297 Norm (lambda vector) = 226.620365 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12427 12118 97.51 97.55 97.53 e-pp 4811 4849 4718 97.30 98.07 97.68 i-np 14376 14426 13985 96.94 97.28 97.11 i-vp 2646 2643 2526 95.57 95.46 95.52 e-vp 4658 4695 4510 96.06 96.82 96.44 e-sbar 535 526 474 90.11 88.60 89.35 o 6180 6182 5973 96.62 96.65 96.63 e-adjp 438 403 349 86.60 79.68 83.00 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 841 707 84.07 81.64 82.84 i-adjp 167 123 107 86.99 64.07 73.79 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 35 31 88.57 64.58 74.70 e-prt 106 105 80 76.19 75.47 75.83 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.83 75.35 77.05 Avg2. 47375 47375 45653 96.37 96.37 96.37 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12427 11730 94.39 94.43 94.41 pp 4811 4849 4710 97.13 97.90 97.52 vp 4658 4695 4395 93.61 94.35 93.98 sbar 535 526 461 87.64 86.17 86.90 adjp 438 403 330 81.89 75.34 78.48 advp 866 841 696 82.76 80.37 81.55 prt 106 105 80 76.19 75.47 75.83 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.61 70.96 74.14 Avg2. 23852 23855 22408 93.93 93.95 93.94 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 76 Log-likelihood = -12755.761965 Norm (log-likelihood gradient vector) = 481.210289 Norm (lambda vector) = 227.869334 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12309 12053 97.92 97.03 97.47 e-pp 4811 4820 4704 97.59 97.78 97.68 i-np 14376 14689 14081 95.86 97.95 96.89 i-vp 2646 2639 2520 95.49 95.24 95.36 e-vp 4658 4698 4505 95.89 96.72 96.30 e-sbar 535 537 477 88.83 89.16 88.99 o 6180 6073 5907 97.27 95.58 96.42 e-adjp 438 391 342 87.47 78.08 82.51 i-advp 89 83 56 67.47 62.92 65.12 e-advp 866 832 705 84.74 81.41 83.04 i-adjp 167 122 107 87.70 64.07 74.05 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 34 31 91.18 64.58 75.61 e-prt 106 106 82 77.36 77.36 77.36 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.79 75.30 76.52 Avg2. 47375 47375 45589 96.23 96.23 96.23 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12309 11631 94.49 93.63 94.06 pp 4811 4820 4696 97.43 97.61 97.52 vp 4658 4698 4387 93.38 94.18 93.78 sbar 535 537 464 86.41 86.73 86.57 adjp 438 391 325 83.12 74.20 78.41 advp 866 832 693 83.29 80.02 81.63 prt 106 106 82 77.36 77.36 77.36 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.55 70.93 73.63 Avg2. 23852 23704 22284 94.01 93.43 93.72 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 77 Log-likelihood = -12597.641096 Norm (log-likelihood gradient vector) = 2532.679043 Norm (lambda vector) = 233.885435 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12394 12100 97.63 97.41 97.52 e-pp 4811 4838 4710 97.35 97.90 97.63 i-np 14376 14501 14009 96.61 97.45 97.03 i-vp 2646 2642 2525 95.57 95.43 95.50 e-vp 4658 4700 4511 95.98 96.84 96.41 e-sbar 535 530 473 89.25 88.41 88.83 o 6180 6148 5953 96.83 96.33 96.58 e-adjp 438 398 347 87.19 79.22 83.01 i-advp 89 83 56 67.47 62.92 65.12 e-advp 866 838 707 84.37 81.64 82.98 i-adjp 167 122 107 87.70 64.07 74.05 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 34 31 91.18 64.58 75.61 e-prt 106 105 81 77.14 76.42 76.78 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.76 75.34 76.53 Avg2. 47375 47375 45629 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12394 11691 94.33 94.12 94.22 pp 4811 4838 4702 97.19 97.73 97.46 vp 4658 4700 4395 93.51 94.35 93.93 sbar 535 530 460 86.79 85.98 86.38 adjp 438 398 329 82.66 75.11 78.71 advp 866 838 695 82.94 80.25 81.57 prt 106 105 81 77.14 76.42 76.78 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.46 70.95 73.60 Avg2. 23852 23814 22359 93.89 93.74 93.82 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 78 Log-likelihood = -12465.412143 Norm (log-likelihood gradient vector) = 928.247700 Norm (lambda vector) = 230.146166 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12404 12105 97.59 97.45 97.52 e-pp 4811 4853 4718 97.22 98.07 97.64 i-np 14376 14472 13996 96.71 97.36 97.03 i-vp 2646 2638 2521 95.56 95.28 95.42 e-vp 4658 4703 4510 95.90 96.82 96.36 e-sbar 535 518 469 90.54 87.66 89.08 o 6180 6163 5960 96.71 96.44 96.57 e-adjp 438 397 348 87.66 79.45 83.35 i-advp 89 83 56 67.47 62.92 65.12 e-advp 866 839 707 84.27 81.64 82.93 i-adjp 167 124 108 87.10 64.67 74.23 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 34 31 91.18 64.58 75.61 e-prt 106 105 81 77.14 76.42 76.78 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.80 75.35 76.55 Avg2. 47375 47375 45629 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12404 11699 94.32 94.18 94.25 pp 4811 4853 4710 97.05 97.90 97.48 vp 4658 4703 4392 93.39 94.29 93.84 sbar 535 518 456 88.03 85.23 86.61 adjp 438 397 330 83.12 75.34 79.04 advp 866 839 695 82.84 80.25 81.52 prt 106 105 81 77.14 76.42 76.78 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.59 70.92 73.64 Avg2. 23852 23830 22369 93.87 93.78 93.83 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 79 Log-likelihood = -12084.393017 Norm (log-likelihood gradient vector) = 531.060711 Norm (lambda vector) = 233.120230 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12417 12111 97.54 97.50 97.52 e-pp 4811 4859 4720 97.14 98.11 97.62 i-np 14376 14447 13983 96.79 97.27 97.03 i-vp 2646 2642 2521 95.42 95.28 95.35 e-vp 4658 4706 4511 95.86 96.84 96.35 e-sbar 535 519 469 90.37 87.66 88.99 o 6180 6167 5961 96.66 96.46 96.56 e-adjp 438 398 347 87.19 79.22 83.01 i-advp 89 81 56 69.14 62.92 65.88 e-advp 866 835 706 84.55 81.52 83.01 i-adjp 167 124 108 87.10 64.67 74.23 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 34 31 91.18 64.58 75.61 e-prt 106 104 81 77.88 76.42 77.14 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.90 75.33 76.59 Avg2. 47375 47375 45624 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12417 11707 94.28 94.24 94.26 pp 4811 4859 4712 96.97 97.94 97.46 vp 4658 4706 4390 93.29 94.25 93.76 sbar 535 519 456 87.86 85.23 86.53 adjp 438 398 328 82.41 74.89 78.47 advp 866 835 694 83.11 80.14 81.60 prt 106 104 81 77.88 76.42 77.14 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.58 70.87 73.61 Avg2. 23852 23849 22374 93.82 93.80 93.81 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 80 Log-likelihood = -11807.678990 Norm (log-likelihood gradient vector) = 350.215792 Norm (lambda vector) = 235.241298 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12428 12116 97.49 97.54 97.51 e-pp 4811 4866 4728 97.16 98.27 97.72 i-np 14376 14424 13973 96.87 97.20 97.03 i-vp 2646 2643 2522 95.42 95.31 95.37 e-vp 4658 4703 4506 95.81 96.74 96.27 e-sbar 535 510 468 91.76 87.48 89.57 o 6180 6171 5964 96.65 96.50 96.58 e-adjp 438 402 349 86.82 79.68 83.10 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 839 709 84.51 81.87 83.17 i-adjp 167 127 111 87.40 66.47 75.51 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 34 31 91.18 64.58 75.61 e-prt 106 104 81 77.88 76.42 77.14 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.93 75.47 76.68 Avg2. 47375 47375 45633 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12428 11712 94.24 94.28 94.26 pp 4811 4866 4720 97.00 98.11 97.55 vp 4658 4703 4384 93.22 94.12 93.67 sbar 535 510 455 89.22 85.05 87.08 adjp 438 402 330 82.09 75.34 78.57 advp 866 839 697 83.08 80.48 81.76 prt 106 104 81 77.88 76.42 77.14 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.67 70.94 73.69 Avg2. 23852 23863 22385 93.81 93.85 93.83 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 81 Log-likelihood = -11476.018107 Norm (log-likelihood gradient vector) = 347.880692 Norm (lambda vector) = 238.245859 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12457 12128 97.36 97.63 97.50 e-pp 4811 4857 4723 97.24 98.17 97.70 i-np 14376 14379 13951 97.02 97.04 97.03 i-vp 2646 2642 2521 95.42 95.28 95.35 e-vp 4658 4698 4504 95.87 96.69 96.28 e-sbar 535 519 470 90.56 87.85 89.18 o 6180 6189 5971 96.48 96.62 96.55 e-adjp 438 404 350 86.63 79.91 83.14 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 839 709 84.51 81.87 83.17 i-adjp 167 127 110 86.61 65.87 74.83 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 35 31 88.57 64.58 74.70 e-prt 106 105 82 78.10 77.36 77.73 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.67 75.52 76.58 Avg2. 47375 47375 45625 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12457 11723 94.11 94.37 94.24 pp 4811 4857 4714 97.06 97.98 97.52 vp 4658 4698 4381 93.25 94.05 93.65 sbar 535 519 457 88.05 85.42 86.72 adjp 438 404 330 81.68 75.34 78.38 advp 866 839 697 83.08 80.48 81.76 prt 106 105 82 78.10 77.36 77.73 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.53 71.06 73.69 Avg2. 23852 23890 22390 93.72 93.87 93.80 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 82 Log-likelihood = -11219.682769 Norm (log-likelihood gradient vector) = 585.268779 Norm (lambda vector) = 240.965276 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12442 12119 97.40 97.56 97.48 e-pp 4811 4864 4724 97.12 98.19 97.65 i-np 14376 14408 13968 96.95 97.16 97.05 i-vp 2646 2640 2521 95.49 95.28 95.38 e-vp 4658 4688 4497 95.93 96.54 96.23 e-sbar 535 516 469 90.89 87.66 89.25 o 6180 6178 5968 96.60 96.57 96.59 e-adjp 438 403 349 86.60 79.68 83.00 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 845 711 84.14 82.10 83.11 i-adjp 167 125 107 85.60 64.07 73.29 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 35 31 88.57 64.58 74.70 e-prt 106 107 83 77.57 78.30 77.93 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 10 71.43 76.92 74.07 e-conjp 9 9 5 55.56 55.56 55.56 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.14 75.51 76.80 Avg2. 47375 47375 45623 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12442 11714 94.15 94.30 94.22 pp 4811 4864 4715 96.94 98.00 97.47 vp 4658 4688 4376 93.34 93.95 93.64 sbar 535 516 456 88.37 85.23 86.77 adjp 438 403 328 81.39 74.89 78.00 advp 866 845 700 82.84 80.83 81.82 prt 106 107 83 77.57 78.30 77.93 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 5 55.56 55.56 55.56 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.02 71.11 73.94 Avg2. 23852 23875 22378 93.73 93.82 93.78 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 83 Log-likelihood = -11012.894745 Norm (log-likelihood gradient vector) = 383.880163 Norm (lambda vector) = 242.914528 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12448 12119 97.36 97.56 97.46 e-pp 4811 4857 4719 97.16 98.09 97.62 i-np 14376 14388 13953 96.98 97.06 97.02 i-vp 2646 2637 2522 95.64 95.31 95.48 e-vp 4658 4693 4500 95.89 96.61 96.25 e-sbar 535 522 471 90.23 88.04 89.12 o 6180 6191 5975 96.51 96.68 96.60 e-adjp 438 404 349 86.39 79.68 82.90 i-advp 89 82 57 69.51 64.04 66.67 e-advp 866 846 713 84.28 82.33 83.29 i-adjp 167 125 108 86.40 64.67 73.97 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 36 31 86.11 64.58 73.81 e-prt 106 106 82 77.36 77.36 77.36 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 10 71.43 76.92 74.07 e-conjp 9 9 5 55.56 55.56 55.56 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.09 75.53 76.79 Avg2. 47375 47375 45618 96.29 96.29 96.29 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12448 11713 94.10 94.29 94.19 pp 4811 4857 4710 96.97 97.90 97.43 vp 4658 4693 4383 93.39 94.10 93.74 sbar 535 522 458 87.74 85.61 86.66 adjp 438 404 328 81.19 74.89 77.91 advp 866 846 703 83.10 81.18 82.13 prt 106 106 82 77.36 77.36 77.36 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 5 55.56 55.56 55.56 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.94 71.09 73.90 Avg2. 23852 23886 22383 93.71 93.84 93.77 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 84 Log-likelihood = -10863.074480 Norm (log-likelihood gradient vector) = 344.448268 Norm (lambda vector) = 244.327889 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12420 12107 97.48 97.46 97.47 e-pp 4811 4864 4723 97.10 98.17 97.63 i-np 14376 14450 13989 96.81 97.31 97.06 i-vp 2646 2635 2521 95.67 95.28 95.47 e-vp 4658 4694 4501 95.89 96.63 96.26 e-sbar 535 516 469 90.89 87.66 89.25 o 6180 6165 5964 96.74 96.50 96.62 e-adjp 438 400 346 86.50 79.00 82.58 i-advp 89 81 57 70.37 64.04 67.06 e-advp 866 847 715 84.42 82.56 83.48 i-adjp 167 120 104 86.67 62.28 72.47 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 31 79.49 64.58 71.26 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 8 66.67 61.54 64.00 e-conjp 9 8 4 50.00 44.44 47.06 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.33 73.98 75.62 Avg2. 47375 47375 45627 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12420 11711 94.29 94.28 94.28 pp 4811 4864 4714 96.92 97.98 97.45 vp 4658 4694 4386 93.44 94.16 93.80 sbar 535 516 456 88.37 85.23 86.77 adjp 438 400 324 81.00 73.97 77.33 advp 866 847 705 83.23 81.41 82.31 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 4 50.00 44.44 47.06 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.58 70.07 73.18 Avg2. 23852 23857 22385 93.83 93.85 93.84 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 85 Log-likelihood = -10711.182064 Norm (log-likelihood gradient vector) = 412.304844 Norm (lambda vector) = 245.400168 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12417 12107 97.50 97.46 97.48 e-pp 4811 4855 4717 97.16 98.05 97.60 i-np 14376 14433 13985 96.90 97.28 97.09 i-vp 2646 2640 2524 95.61 95.39 95.50 e-vp 4658 4694 4504 95.95 96.69 96.32 e-sbar 535 526 473 89.92 88.41 89.16 o 6180 6173 5969 96.70 96.59 96.64 e-adjp 438 406 348 85.71 79.45 82.46 i-advp 89 79 56 70.89 62.92 66.67 e-advp 866 850 714 84.00 82.45 83.22 i-adjp 167 121 106 87.60 63.47 73.61 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 31 79.49 64.58 71.26 e-prt 106 105 82 78.10 77.36 77.73 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 8 66.67 61.54 64.00 e-conjp 9 8 4 50.00 44.44 47.06 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.27 73.95 75.58 Avg2. 47375 47375 45632 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12417 11711 94.31 94.28 94.30 pp 4811 4855 4708 96.97 97.86 97.41 vp 4658 4694 4388 93.48 94.20 93.84 sbar 535 526 460 87.45 85.98 86.71 adjp 438 406 326 80.30 74.43 77.25 advp 866 850 704 82.82 81.29 82.05 prt 106 105 82 78.10 77.36 77.73 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 4 50.00 44.44 47.06 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.34 69.98 73.03 Avg2. 23852 23862 22384 93.81 93.85 93.83 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 86 Log-likelihood = -10526.313143 Norm (log-likelihood gradient vector) = 317.005415 Norm (lambda vector) = 247.163312 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12409 12099 97.50 97.40 97.45 e-pp 4811 4848 4714 97.24 97.98 97.61 i-np 14376 14460 13988 96.74 97.30 97.02 i-vp 2646 2652 2529 95.36 95.58 95.47 e-vp 4658 4700 4506 95.87 96.74 96.30 e-sbar 535 525 471 89.71 88.04 88.87 o 6180 6157 5955 96.72 96.36 96.54 e-adjp 438 400 349 87.25 79.68 83.29 i-advp 89 78 56 71.79 62.92 67.07 e-advp 866 837 711 84.95 82.10 83.50 i-adjp 167 124 109 87.90 65.27 74.91 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 32 86.49 66.67 75.29 e-prt 106 106 83 78.30 78.30 78.30 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.85 75.66 76.74 Avg2. 47375 47375 45621 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12409 11694 94.24 94.14 94.19 pp 4811 4848 4705 97.05 97.80 97.42 vp 4658 4700 4387 93.34 94.18 93.76 sbar 535 525 458 87.24 85.61 86.42 adjp 438 400 330 82.50 75.34 78.76 advp 866 837 700 83.63 80.83 82.21 prt 106 106 83 78.30 78.30 78.30 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.63 71.18 73.80 Avg2. 23852 23836 22363 93.82 93.76 93.79 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 87 Log-likelihood = -10246.560554 Norm (log-likelihood gradient vector) = 637.654881 Norm (lambda vector) = 251.328588 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12458 12119 97.28 97.56 97.42 e-pp 4811 4871 4732 97.15 98.36 97.75 i-np 14376 14364 13933 97.00 96.92 96.96 i-vp 2646 2653 2530 95.36 95.62 95.49 e-vp 4658 4699 4503 95.83 96.67 96.25 e-sbar 535 508 467 91.93 87.29 89.55 o 6180 6190 5970 96.45 96.60 96.52 e-adjp 438 404 351 86.88 80.14 83.37 i-advp 89 79 56 70.89 62.92 66.67 e-advp 866 840 710 84.52 81.99 83.24 i-adjp 167 126 111 88.10 66.47 75.77 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 103 81 78.64 76.42 77.51 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.76 75.62 76.67 Avg2. 47375 47375 45614 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12458 11700 93.92 94.19 94.05 pp 4811 4871 4722 96.94 98.15 97.54 vp 4658 4699 4384 93.30 94.12 93.71 sbar 535 508 454 89.37 84.86 87.06 adjp 438 404 331 81.93 75.57 78.62 advp 866 840 699 83.21 80.72 81.95 prt 106 103 81 78.64 76.42 77.51 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.73 70.96 73.73 Avg2. 23852 23894 22377 93.65 93.82 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 88 Log-likelihood = -10128.625625 Norm (log-likelihood gradient vector) = 555.265006 Norm (lambda vector) = 253.723330 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12442 12118 97.40 97.55 97.47 e-pp 4811 4862 4725 97.18 98.21 97.69 i-np 14376 14396 13959 96.96 97.10 97.03 i-vp 2646 2653 2530 95.36 95.62 95.49 e-vp 4658 4695 4503 95.91 96.67 96.29 e-sbar 535 515 469 91.07 87.66 89.33 o 6180 6180 5968 96.57 96.57 96.57 e-adjp 438 403 351 87.10 80.14 83.47 i-advp 89 79 56 70.89 62.92 66.67 e-advp 866 839 710 84.62 81.99 83.28 i-adjp 167 126 111 88.10 66.47 75.77 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 105 82 78.10 77.36 77.73 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.72 75.69 76.69 Avg2. 47375 47375 45633 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12442 11710 94.12 94.27 94.19 pp 4811 4862 4715 96.98 98.00 97.49 vp 4658 4695 4383 93.35 94.10 93.72 sbar 535 515 456 88.54 85.23 86.86 adjp 438 403 331 82.13 75.57 78.72 advp 866 839 699 83.31 80.72 81.99 prt 106 105 82 78.10 77.36 77.73 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.65 71.08 73.76 Avg2. 23852 23872 22382 93.76 93.84 93.80 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 89 Log-likelihood = -10042.137467 Norm (log-likelihood gradient vector) = 342.599462 Norm (lambda vector) = 253.295446 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12431 12112 97.43 97.50 97.47 e-pp 4811 4856 4720 97.20 98.11 97.65 i-np 14376 14422 13970 96.87 97.18 97.02 i-vp 2646 2647 2526 95.43 95.46 95.45 e-vp 4658 4695 4506 95.97 96.74 96.35 e-sbar 535 521 471 90.40 88.04 89.20 o 6180 6170 5962 96.63 96.47 96.55 e-adjp 438 402 350 87.06 79.91 83.33 i-advp 89 79 56 70.89 62.92 66.67 e-advp 866 838 710 84.73 81.99 83.33 i-adjp 167 128 111 86.72 66.47 75.25 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 32 86.49 66.67 75.29 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.76 75.79 76.76 Avg2. 47375 47375 45629 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12431 11704 94.15 94.22 94.19 pp 4811 4856 4711 97.01 97.92 97.47 vp 4658 4695 4385 93.40 94.14 93.77 sbar 535 521 458 87.91 85.61 86.74 adjp 438 402 331 82.34 75.57 78.81 advp 866 838 699 83.41 80.72 82.04 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.67 71.30 73.89 Avg2. 23852 23861 22378 93.78 93.82 93.80 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 90 Log-likelihood = -9874.694822 Norm (log-likelihood gradient vector) = 210.970912 Norm (lambda vector) = 254.073635 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12422 12106 97.46 97.46 97.46 e-pp 4811 4852 4718 97.24 98.07 97.65 i-np 14376 14440 13983 96.84 97.27 97.05 i-vp 2646 2648 2526 95.39 95.46 95.43 e-vp 4658 4693 4504 95.97 96.69 96.33 e-sbar 535 524 472 90.08 88.22 89.14 o 6180 6162 5961 96.74 96.46 96.60 e-adjp 438 403 349 86.60 79.68 83.00 i-advp 89 79 56 70.89 62.92 66.67 e-advp 866 839 710 84.62 81.99 83.28 i-adjp 167 128 111 86.72 66.47 75.25 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 32 86.49 66.67 75.29 e-prt 106 106 84 79.25 79.25 79.25 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.76 75.78 76.76 Avg2. 47375 47375 45631 96.32 96.32 96.32 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12422 11700 94.19 94.19 94.19 pp 4811 4852 4709 97.05 97.88 97.46 vp 4658 4693 4385 93.44 94.14 93.79 sbar 535 524 459 87.60 85.79 86.69 adjp 438 403 329 81.64 75.11 78.24 advp 866 839 699 83.31 80.72 81.99 prt 106 106 84 79.25 79.25 79.25 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.65 71.26 73.86 Avg2. 23852 23850 22371 93.80 93.79 93.79 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 91 Log-likelihood = -9688.982498 Norm (log-likelihood gradient vector) = 295.197475 Norm (lambda vector) = 256.373267 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12421 12102 97.43 97.42 97.43 e-pp 4811 4844 4712 97.27 97.94 97.61 i-np 14376 14440 13976 96.79 97.22 97.00 i-vp 2646 2651 2526 95.28 95.46 95.37 e-vp 4658 4689 4500 95.97 96.61 96.29 e-sbar 535 531 476 89.64 88.97 89.31 o 6180 6167 5958 96.61 96.41 96.51 e-adjp 438 401 346 86.28 79.00 82.48 i-advp 89 80 56 70.00 62.92 66.27 e-advp 866 842 711 84.44 82.10 83.26 i-adjp 167 124 107 86.29 64.07 73.54 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 32 86.49 66.67 75.29 e-prt 106 106 84 79.25 79.25 79.25 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.62 75.64 76.62 Avg2. 47375 47375 45605 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12421 11696 94.16 94.16 94.16 pp 4811 4844 4703 97.09 97.76 97.42 vp 4658 4689 4383 93.47 94.10 93.78 sbar 535 531 463 87.19 86.54 86.87 adjp 438 401 324 80.80 73.97 77.23 advp 866 842 700 83.14 80.83 81.97 prt 106 106 84 79.25 79.25 79.25 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.51 71.22 73.77 Avg2. 23852 23845 22359 93.77 93.74 93.75 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 92 Log-likelihood = -9441.916011 Norm (log-likelihood gradient vector) = 319.746066 Norm (lambda vector) = 260.496801 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12295 12018 97.75 96.75 97.24 e-pp 4811 4879 4734 97.03 98.40 97.71 i-np 14376 14713 14084 95.72 97.97 96.83 i-vp 2646 2648 2521 95.20 95.28 95.24 e-vp 4658 4684 4493 95.92 96.46 96.19 e-sbar 535 494 459 92.91 85.79 89.21 o 6180 6034 5880 97.45 95.15 96.28 e-adjp 438 401 345 86.03 78.77 82.24 i-advp 89 80 57 71.25 64.04 67.46 e-advp 866 844 708 83.89 81.76 82.81 i-adjp 167 116 102 87.93 61.08 72.08 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 40 33 82.50 68.75 75.00 e-prt 106 106 83 78.30 78.30 78.30 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 8 57.14 61.54 59.26 e-conjp 9 10 4 40.00 44.44 42.11 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.54 73.86 75.18 Avg2. 47375 47375 45533 96.11 96.11 96.11 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12295 11598 94.33 93.37 93.85 pp 4811 4879 4727 96.88 98.25 97.56 vp 4658 4684 4371 93.32 93.84 93.58 sbar 535 494 446 90.28 83.36 86.69 adjp 438 401 322 80.30 73.52 76.76 advp 866 844 697 82.58 80.48 81.52 prt 106 106 83 78.30 78.30 78.30 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 4 40.00 44.44 42.11 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.60 69.56 72.45 Avg2. 23852 23714 22249 93.82 93.28 93.55 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 93 Log-likelihood = -10063.575877 Norm (log-likelihood gradient vector) = 2707.176408 Norm (lambda vector) = 269.978198 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12391 12085 97.53 97.29 97.41 e-pp 4811 4858 4721 97.18 98.13 97.65 i-np 14376 14495 14004 96.61 97.41 97.01 i-vp 2646 2649 2523 95.24 95.35 95.30 e-vp 4658 4687 4497 95.95 96.54 96.24 e-sbar 535 520 472 90.77 88.22 89.48 o 6180 6140 5947 96.86 96.23 96.54 e-adjp 438 400 345 86.25 78.77 82.34 i-advp 89 82 57 69.51 64.04 66.67 e-advp 866 844 712 84.36 82.22 83.27 i-adjp 167 121 105 86.78 62.87 72.92 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 32 86.49 66.67 75.29 e-prt 106 107 85 79.44 80.19 79.81 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 16 10 62.50 76.92 68.97 e-conjp 9 11 5 45.45 55.56 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.20 75.63 76.41 Avg2. 47375 47375 45604 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12391 11678 94.25 94.01 94.13 pp 4811 4858 4713 97.02 97.96 97.49 vp 4658 4687 4377 93.39 93.97 93.68 sbar 535 520 459 88.27 85.79 87.01 adjp 438 400 323 80.75 73.74 77.09 advp 866 844 701 83.06 80.95 81.99 prt 106 107 85 79.44 80.19 79.81 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 11 5 45.45 55.56 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.16 71.22 73.61 Avg2. 23852 23819 22342 93.80 93.67 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 94 Log-likelihood = -9346.357025 Norm (log-likelihood gradient vector) = 593.146100 Norm (lambda vector) = 263.104684 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12409 12099 97.50 97.40 97.45 e-pp 4811 4851 4716 97.22 98.03 97.62 i-np 14376 14456 13992 96.79 97.33 97.06 i-vp 2646 2648 2520 95.17 95.24 95.20 e-vp 4658 4691 4496 95.84 96.52 96.18 e-sbar 535 523 472 90.25 88.22 89.22 o 6180 6158 5957 96.74 96.39 96.56 e-adjp 438 402 345 85.82 78.77 82.14 i-advp 89 81 57 70.37 64.04 67.06 e-advp 866 848 712 83.96 82.22 83.08 i-adjp 167 120 104 86.67 62.28 72.47 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 37 32 86.49 66.67 75.29 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 16 10 62.50 76.92 68.97 e-conjp 9 11 5 45.45 55.56 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.11 75.55 76.32 Avg2. 47375 47375 45605 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12409 11700 94.29 94.19 94.24 pp 4811 4851 4708 97.05 97.86 97.45 vp 4658 4691 4375 93.26 93.92 93.59 sbar 535 523 459 87.76 85.79 86.77 adjp 438 402 323 80.35 73.74 76.90 advp 866 848 701 82.67 80.95 81.80 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 11 5 45.45 55.56 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.93 71.13 73.45 Avg2. 23852 23843 22356 93.76 93.73 93.75 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 95 Log-likelihood = -9164.292733 Norm (log-likelihood gradient vector) = 354.159711 Norm (lambda vector) = 267.316096 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12428 12109 97.43 97.48 97.46 e-pp 4811 4856 4719 97.18 98.09 97.63 i-np 14376 14427 13984 96.93 97.27 97.10 i-vp 2646 2650 2522 95.17 95.31 95.24 e-vp 4658 4690 4496 95.86 96.52 96.19 e-sbar 535 521 471 90.40 88.04 89.20 o 6180 6172 5965 96.65 96.52 96.58 e-adjp 438 400 342 85.50 78.08 81.62 i-advp 89 82 57 69.51 64.04 66.67 e-advp 866 850 712 83.76 82.22 82.98 i-adjp 167 113 101 89.38 60.48 72.14 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 41 32 78.05 66.67 71.91 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 12 6 50.00 46.15 48.00 e-conjp 9 9 3 33.33 33.33 33.33 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.36 72.47 73.89 Avg2. 47375 47375 45607 96.27 96.27 96.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12428 11714 94.25 94.30 94.28 pp 4811 4856 4711 97.01 97.92 97.47 vp 4658 4690 4376 93.30 93.95 93.62 sbar 535 521 458 87.91 85.61 86.74 adjp 438 400 320 80.00 73.06 76.37 advp 866 850 700 82.35 80.83 81.59 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 3 33.33 33.33 33.33 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.67 68.82 71.63 Avg2. 23852 23862 22367 93.73 93.77 93.75 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 96 Log-likelihood = -9025.331941 Norm (log-likelihood gradient vector) = 254.930157 Norm (lambda vector) = 269.694051 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12467 12120 97.22 97.57 97.39 e-pp 4811 4858 4718 97.12 98.07 97.59 i-np 14376 14337 13926 97.13 96.87 97.00 i-vp 2646 2655 2524 95.07 95.39 95.23 e-vp 4658 4701 4498 95.68 96.57 96.12 e-sbar 535 527 471 89.37 88.04 88.70 o 6180 6203 5974 96.31 96.67 96.49 e-adjp 438 398 343 86.18 78.31 82.06 i-advp 89 85 57 67.06 64.04 65.52 e-advp 866 850 708 83.29 81.76 82.52 i-adjp 167 113 101 89.38 60.48 72.14 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 43 32 74.42 66.67 70.33 e-prt 106 105 81 77.14 76.42 76.78 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 9 4 44.44 30.77 36.36 e-conjp 9 7 2 28.57 22.22 25.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.29 70.82 72.51 Avg2. 47375 47375 45563 96.18 96.18 96.18 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12467 11709 93.92 94.26 94.09 pp 4811 4858 4707 96.89 97.84 97.36 vp 4658 4701 4374 93.04 93.90 93.47 sbar 535 527 458 86.91 85.61 86.25 adjp 438 398 322 80.90 73.52 77.03 advp 866 850 696 81.88 80.37 81.12 prt 106 105 81 77.14 76.42 76.78 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 2 28.57 22.22 25.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 73.93 67.41 70.52 Avg2. 23852 23914 22350 93.46 93.70 93.58 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 97 Log-likelihood = -8941.309536 Norm (log-likelihood gradient vector) = 672.587831 Norm (lambda vector) = 272.136737 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12439 12116 97.40 97.54 97.47 e-pp 4811 4856 4719 97.18 98.09 97.63 i-np 14376 14399 13968 97.01 97.16 97.08 i-vp 2646 2661 2530 95.08 95.62 95.35 e-vp 4658 4696 4500 95.83 96.61 96.22 e-sbar 535 527 471 89.37 88.04 88.70 o 6180 6180 5968 96.57 96.57 96.57 e-adjp 438 395 344 87.09 78.54 82.59 i-advp 89 83 56 67.47 62.92 65.12 e-advp 866 844 705 83.53 81.41 82.46 i-adjp 167 114 102 89.47 61.08 72.60 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 43 32 74.42 66.67 70.33 e-prt 106 105 82 78.10 77.36 77.73 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 9 4 44.44 30.77 36.36 e-conjp 9 7 2 28.57 22.22 25.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.46 70.87 72.62 Avg2. 47375 47375 45603 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12439 11715 94.18 94.31 94.24 pp 4811 4856 4708 96.95 97.86 97.40 vp 4658 4696 4380 93.27 94.03 93.65 sbar 535 527 458 86.91 85.61 86.25 adjp 438 395 322 81.52 73.52 77.31 advp 866 844 693 82.11 80.02 81.05 prt 106 105 82 78.10 77.36 77.73 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 2 28.57 22.22 25.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 74.16 67.49 70.67 Avg2. 23852 23870 22361 93.68 93.75 93.71 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 98 Log-likelihood = -8803.180959 Norm (log-likelihood gradient vector) = 265.213827 Norm (lambda vector) = 272.080450 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12429 12110 97.43 97.49 97.46 e-pp 4811 4858 4722 97.20 98.15 97.67 i-np 14376 14428 13984 96.92 97.27 97.10 i-vp 2646 2660 2529 95.08 95.58 95.33 e-vp 4658 4697 4501 95.83 96.63 96.23 e-sbar 535 521 468 89.83 87.48 88.64 o 6180 6166 5961 96.68 96.46 96.57 e-adjp 438 397 345 86.90 78.77 82.63 i-advp 89 81 56 69.14 62.92 65.88 e-advp 866 843 707 83.87 81.64 82.74 i-adjp 167 114 102 89.47 61.08 72.60 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 105 82 78.10 77.36 77.73 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 6 54.55 46.15 50.00 e-conjp 9 8 3 37.50 33.33 35.29 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.96 72.33 74.10 Avg2. 47375 47375 45612 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12429 11712 94.23 94.28 94.26 pp 4811 4858 4711 96.97 97.92 97.45 vp 4658 4697 4379 93.23 94.01 93.62 sbar 535 521 455 87.33 85.05 86.17 adjp 438 397 322 81.11 73.52 77.13 advp 866 843 695 82.44 80.25 81.33 prt 106 105 82 78.10 77.36 77.73 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 3 37.50 33.33 35.29 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.09 68.57 71.68 Avg2. 23852 23859 22360 93.72 93.74 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 99 Log-likelihood = -8744.409220 Norm (log-likelihood gradient vector) = 210.375249 Norm (lambda vector) = 271.058103 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12412 12102 97.50 97.42 97.46 e-pp 4811 4849 4720 97.34 98.11 97.72 i-np 14376 14456 13996 96.82 97.36 97.09 i-vp 2646 2661 2529 95.04 95.58 95.31 e-vp 4658 4696 4500 95.83 96.61 96.22 e-sbar 535 527 472 89.56 88.22 88.89 o 6180 6149 5956 96.86 96.38 96.62 e-adjp 438 400 346 86.50 79.00 82.58 i-advp 89 81 56 69.14 62.92 65.88 e-advp 866 836 707 84.57 81.64 83.08 i-adjp 167 123 107 86.99 64.07 73.79 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 8 61.54 61.54 61.54 e-conjp 9 9 4 44.44 44.44 44.44 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.89 74.24 75.54 Avg2. 47375 47375 45624 96.30 96.30 96.30 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12412 11700 94.26 94.19 94.23 pp 4811 4849 4710 97.13 97.90 97.52 vp 4658 4696 4379 93.25 94.01 93.63 sbar 535 527 459 87.10 85.79 86.44 adjp 438 400 326 81.50 74.43 77.80 advp 866 836 695 83.13 80.25 81.67 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 4 44.44 44.44 44.44 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.93 70.03 72.86 Avg2. 23852 23837 22358 93.80 93.74 93.77 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 100 Log-likelihood = -8659.175253 Norm (log-likelihood gradient vector) = 298.929050 Norm (lambda vector) = 270.998457 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12430 12108 97.41 97.47 97.44 e-pp 4811 4863 4726 97.18 98.23 97.71 i-np 14376 14431 13982 96.89 97.26 97.07 i-vp 2646 2650 2524 95.25 95.39 95.32 e-vp 4658 4694 4502 95.91 96.65 96.28 e-sbar 535 513 465 90.64 86.92 88.74 o 6180 6165 5962 96.71 96.47 96.59 e-adjp 438 402 349 86.82 79.68 83.10 i-advp 89 80 56 70.00 62.92 66.27 e-advp 866 831 705 84.84 81.41 83.09 i-adjp 167 126 110 87.30 65.87 75.09 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 38 33 86.84 68.75 76.74 e-prt 106 110 86 78.18 81.13 79.63 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 15 10 66.67 76.92 71.43 e-conjp 9 10 5 50.00 55.56 52.63 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.74 75.87 76.79 Avg2. 47375 47375 45627 96.31 96.31 96.31 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12430 11709 94.20 94.26 94.23 pp 4811 4863 4717 97.00 98.05 97.52 vp 4658 4694 4381 93.33 94.05 93.69 sbar 535 513 452 88.11 84.49 86.26 adjp 438 402 329 81.84 75.11 78.33 advp 866 831 693 83.39 80.02 81.67 prt 106 110 86 78.18 81.13 79.63 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 10 5 50.00 55.56 52.63 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.61 71.27 73.84 Avg2. 23852 23854 22373 93.79 93.80 93.80 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 101 Log-likelihood = -8571.603782 Norm (log-likelihood gradient vector) = 289.334933 Norm (lambda vector) = 271.370266 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12430 12104 97.38 97.44 97.41 e-pp 4811 4859 4722 97.18 98.15 97.66 i-np 14376 14437 13982 96.85 97.26 97.05 i-vp 2646 2652 2525 95.21 95.43 95.32 e-vp 4658 4687 4497 95.95 96.54 96.24 e-sbar 535 516 466 90.31 87.10 88.68 o 6180 6164 5959 96.67 96.42 96.55 e-adjp 438 404 350 86.63 79.91 83.14 i-advp 89 82 56 68.29 62.92 65.50 e-advp 866 829 703 84.80 81.18 82.95 i-adjp 167 127 111 87.40 66.47 75.51 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 33 84.62 68.75 75.86 e-prt 106 109 85 77.98 80.19 79.07 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 10 71.43 76.92 74.07 e-conjp 9 9 5 55.56 55.56 55.56 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.06 75.85 76.94 Avg2. 47375 47375 45612 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12430 11702 94.14 94.20 94.17 pp 4811 4859 4713 97.00 97.96 97.48 vp 4658 4687 4375 93.34 93.92 93.63 sbar 535 516 453 87.79 84.67 86.20 adjp 438 404 332 82.18 75.80 78.86 advp 866 829 691 83.35 79.79 81.53 prt 106 109 85 77.98 80.19 79.07 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 5 55.56 55.56 55.56 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.13 71.21 74.05 Avg2. 23852 23844 22357 93.76 93.73 93.75 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 102 Log-likelihood = -8468.651839 Norm (log-likelihood gradient vector) = 238.542584 Norm (lambda vector) = 273.520575 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12443 12109 97.32 97.48 97.40 e-pp 4811 4856 4719 97.18 98.09 97.63 i-np 14376 14408 13966 96.93 97.15 97.04 i-vp 2646 2654 2528 95.25 95.54 95.40 e-vp 4658 4688 4501 96.01 96.63 96.32 e-sbar 535 518 467 90.15 87.29 88.70 o 6180 6182 5965 96.49 96.52 96.51 e-adjp 438 399 347 86.97 79.22 82.92 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 834 707 84.77 81.64 83.18 i-adjp 167 124 110 88.71 65.87 75.60 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 38 33 86.84 68.75 76.74 e-prt 106 107 85 79.44 80.19 79.81 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 14 10 71.43 76.92 74.07 e-conjp 9 9 5 55.56 55.56 55.56 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.31 75.88 77.08 Avg2. 47375 47375 45613 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12443 11707 94.09 94.24 94.16 pp 4811 4856 4711 97.01 97.92 97.47 vp 4658 4688 4380 93.43 94.03 93.73 sbar 535 518 454 87.64 84.86 86.23 adjp 438 399 330 82.71 75.34 78.85 advp 866 834 695 83.33 80.25 81.76 prt 106 107 85 79.44 80.19 79.81 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 9 5 55.56 55.56 55.56 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.32 71.24 74.16 Avg2. 23852 23855 22368 93.77 93.78 93.77 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 103 Log-likelihood = -8240.392722 Norm (log-likelihood gradient vector) = 189.162748 Norm (lambda vector) = 278.419333 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12405 12090 97.46 97.33 97.39 e-pp 4811 4851 4717 97.24 98.05 97.64 i-np 14376 14482 14012 96.75 97.47 97.11 i-vp 2646 2655 2529 95.25 95.58 95.42 e-vp 4658 4687 4499 95.99 96.59 96.29 e-sbar 535 522 469 89.85 87.66 88.74 o 6180 6143 5952 96.89 96.31 96.60 e-adjp 438 407 350 86.00 79.91 82.84 i-advp 89 85 57 67.06 64.04 65.52 e-advp 866 832 702 84.38 81.06 82.69 i-adjp 167 122 108 88.52 64.67 74.74 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 41 33 80.49 68.75 74.16 e-prt 106 107 85 79.44 80.19 79.81 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.94 74.37 76.11 Avg2. 47375 47375 45619 96.29 96.29 96.29 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12405 11694 94.27 94.14 94.20 pp 4811 4851 4706 97.01 97.82 97.41 vp 4658 4687 4381 93.47 94.05 93.76 sbar 535 522 455 87.16 85.05 86.09 adjp 438 407 330 81.08 75.34 78.11 advp 866 832 689 82.81 79.56 81.15 prt 106 107 85 79.44 80.19 79.81 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.24 70.06 73.47 Avg2. 23852 23819 22345 93.81 93.68 93.75 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 104 Log-likelihood = -8130.562232 Norm (log-likelihood gradient vector) = 409.264556 Norm (lambda vector) = 284.834291 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12449 12110 97.28 97.49 97.38 e-pp 4811 4846 4716 97.32 98.03 97.67 i-np 14376 14393 13958 96.98 97.09 97.04 i-vp 2646 2656 2528 95.18 95.54 95.36 e-vp 4658 4690 4500 95.95 96.61 96.28 e-sbar 535 524 470 89.69 87.85 88.76 o 6180 6187 5969 96.48 96.59 96.53 e-adjp 438 407 351 86.24 80.14 83.08 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 835 704 84.31 81.29 82.77 i-adjp 167 122 108 88.52 64.67 74.74 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 106 84 79.25 79.25 79.25 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.92 74.24 76.03 Avg2. 47375 47375 45603 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12449 11706 94.03 94.24 94.13 pp 4811 4846 4704 97.07 97.78 97.42 vp 4658 4690 4379 93.37 94.01 93.69 sbar 535 524 456 87.02 85.23 86.12 adjp 438 407 331 81.33 75.57 78.34 advp 866 835 691 82.75 79.79 81.25 prt 106 106 84 79.25 79.25 79.25 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.20 70.03 73.44 Avg2. 23852 23865 22356 93.68 93.73 93.70 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 105 Log-likelihood = -8029.188461 Norm (log-likelihood gradient vector) = 323.666899 Norm (lambda vector) = 285.920519 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12438 12104 97.31 97.44 97.38 e-pp 4811 4849 4717 97.28 98.05 97.66 i-np 14376 14412 13964 96.89 97.13 97.01 i-vp 2646 2656 2529 95.22 95.58 95.40 e-vp 4658 4696 4501 95.85 96.63 96.24 e-sbar 535 522 469 89.85 87.66 88.74 o 6180 6180 5966 96.54 96.54 96.54 e-adjp 438 404 349 86.39 79.68 82.90 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 832 704 84.62 81.29 82.92 i-adjp 167 120 106 88.33 63.47 73.87 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 106 84 79.25 79.25 79.25 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.94 74.13 75.99 Avg2. 47375 47375 45598 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12438 11696 94.03 94.16 94.09 pp 4811 4849 4705 97.03 97.80 97.41 vp 4658 4696 4381 93.29 94.05 93.67 sbar 535 522 455 87.16 85.05 86.09 adjp 438 404 329 81.44 75.11 78.15 advp 866 832 691 83.05 79.79 81.39 prt 106 106 84 79.25 79.25 79.25 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.24 69.96 73.42 Avg2. 23852 23855 22346 93.67 93.69 93.68 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 106 Log-likelihood = -7966.588849 Norm (log-likelihood gradient vector) = 174.242458 Norm (lambda vector) = 285.389389 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12434 12106 97.36 97.46 97.41 e-pp 4811 4852 4718 97.24 98.07 97.65 i-np 14376 14410 13967 96.93 97.15 97.04 i-vp 2646 2656 2530 95.26 95.62 95.44 e-vp 4658 4702 4505 95.81 96.72 96.26 e-sbar 535 521 469 90.02 87.66 88.83 o 6180 6176 5968 96.63 96.57 96.60 e-adjp 438 402 348 86.57 79.45 82.86 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 837 705 84.23 81.41 82.80 i-adjp 167 119 106 89.08 63.47 74.13 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 106 84 79.25 79.25 79.25 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.99 74.14 76.01 Avg2. 47375 47375 45611 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12434 11703 94.12 94.21 94.17 pp 4811 4852 4706 96.99 97.82 97.40 vp 4658 4702 4384 93.24 94.12 93.68 sbar 535 521 455 87.33 85.05 86.17 adjp 438 402 328 81.59 74.89 78.10 advp 866 837 692 82.68 79.91 81.27 prt 106 106 84 79.25 79.25 79.25 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.23 69.97 73.42 Avg2. 23852 23862 22357 93.69 93.73 93.71 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 107 Log-likelihood = -7916.039989 Norm (log-likelihood gradient vector) = 192.967790 Norm (lambda vector) = 285.440965 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12425 12103 97.41 97.43 97.42 e-pp 4811 4850 4720 97.32 98.11 97.71 i-np 14376 14429 13976 96.86 97.22 97.04 i-vp 2646 2654 2529 95.29 95.58 95.43 e-vp 4658 4701 4506 95.85 96.74 96.29 e-sbar 535 525 471 89.71 88.04 88.87 o 6180 6163 5961 96.72 96.46 96.59 e-adjp 438 403 349 86.60 79.68 83.00 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 838 706 84.25 81.52 82.86 i-adjp 167 121 107 88.43 64.07 74.31 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.95 74.21 76.04 Avg2. 47375 47375 45617 96.29 96.29 96.29 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12425 11699 94.16 94.18 94.17 pp 4811 4850 4709 97.09 97.88 97.48 vp 4658 4701 4386 93.30 94.16 93.73 sbar 535 525 457 87.05 85.42 86.23 adjp 438 403 329 81.64 75.11 78.24 advp 866 838 694 82.82 80.14 81.46 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.17 70.06 73.44 Avg2. 23852 23857 22363 93.74 93.76 93.75 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 108 Log-likelihood = -7784.105299 Norm (log-likelihood gradient vector) = 258.224471 Norm (lambda vector) = 286.239249 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12456 12112 97.24 97.50 97.37 e-pp 4811 4868 4726 97.08 98.23 97.65 i-np 14376 14376 13944 96.99 96.99 96.99 i-vp 2646 2654 2531 95.37 95.65 95.51 e-vp 4658 4693 4504 95.97 96.69 96.33 e-sbar 535 507 459 90.53 85.79 88.10 o 6180 6188 5970 96.48 96.60 96.54 e-adjp 438 409 349 85.33 79.68 82.41 i-advp 89 82 57 69.51 64.04 66.67 e-advp 866 840 707 84.17 81.64 82.88 i-adjp 167 122 107 87.70 64.07 74.05 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 43 31 72.09 64.58 68.13 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 7 4 57.14 30.77 40.00 e-conjp 9 5 2 40.00 22.22 28.57 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.65 71.04 73.27 Avg2. 47375 47375 45591 96.23 96.23 96.23 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12456 11704 93.96 94.22 94.09 pp 4811 4868 4715 96.86 98.00 97.43 vp 4658 4693 4388 93.50 94.20 93.85 sbar 535 507 445 87.77 83.18 85.41 adjp 438 409 329 80.44 75.11 77.69 advp 866 840 695 82.74 80.25 81.48 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 5 2 40.00 22.22 28.57 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 75.38 67.64 71.30 Avg2. 23852 23886 22363 93.62 93.76 93.69 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 109 Log-likelihood = -7706.716805 Norm (log-likelihood gradient vector) = 579.443960 Norm (lambda vector) = 288.488629 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12430 12105 97.39 97.45 97.42 e-pp 4811 4851 4718 97.26 98.07 97.66 i-np 14376 14420 13971 96.89 97.18 97.03 i-vp 2646 2655 2531 95.33 95.65 95.49 e-vp 4658 4696 4505 95.93 96.72 96.32 e-sbar 535 522 466 89.27 87.10 88.17 o 6180 6170 5963 96.65 96.49 96.57 e-adjp 438 408 349 85.54 79.68 82.51 i-advp 89 81 57 70.37 64.04 67.06 e-advp 866 837 706 84.35 81.52 82.91 i-adjp 167 123 108 87.80 64.67 74.48 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 42 32 76.19 66.67 71.11 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 9 6 66.67 46.15 54.55 e-conjp 9 6 3 50.00 33.33 40.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.99 72.72 74.79 Avg2. 47375 47375 45608 96.27 96.27 96.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12430 11699 94.12 94.18 94.15 pp 4811 4851 4707 97.03 97.84 97.43 vp 4658 4696 4388 93.44 94.20 93.82 sbar 535 522 452 86.59 84.49 85.53 adjp 438 408 330 80.88 75.34 78.01 advp 866 837 695 83.03 80.25 81.62 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 6 3 50.00 33.33 40.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.36 68.89 72.43 Avg2. 23852 23858 22359 93.72 93.74 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 110 Log-likelihood = -7568.486624 Norm (log-likelihood gradient vector) = 198.619991 Norm (lambda vector) = 289.085867 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12423 12103 97.42 97.43 97.43 e-pp 4811 4848 4717 97.30 98.05 97.67 i-np 14376 14427 13973 96.85 97.20 97.02 i-vp 2646 2654 2531 95.37 95.65 95.51 e-vp 4658 4698 4504 95.87 96.69 96.28 e-sbar 535 526 469 89.16 87.66 88.41 o 6180 6167 5962 96.68 96.47 96.57 e-adjp 438 408 348 85.29 79.45 82.27 i-advp 89 82 57 69.51 64.04 66.67 e-advp 866 835 706 84.55 81.52 83.01 i-adjp 167 123 108 87.80 64.67 74.48 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 108 84 77.78 79.25 78.50 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.84 74.21 75.98 Avg2. 47375 47375 45610 96.27 96.27 96.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12423 11697 94.16 94.16 94.16 pp 4811 4848 4706 97.07 97.82 97.44 vp 4658 4698 4388 93.40 94.20 93.80 sbar 535 526 455 86.50 85.05 85.77 adjp 438 408 329 80.64 75.11 77.78 advp 866 835 695 83.23 80.25 81.72 prt 106 108 84 77.78 79.25 78.50 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 76.99 70.03 73.35 Avg2. 23852 23854 22359 93.73 93.74 93.74 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 111 Log-likelihood = -7510.301245 Norm (log-likelihood gradient vector) = 161.005337 Norm (lambda vector) = 289.814634 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12426 12104 97.41 97.44 97.42 e-pp 4811 4839 4714 97.42 97.98 97.70 i-np 14376 14422 13968 96.85 97.16 97.01 i-vp 2646 2651 2528 95.36 95.54 95.45 e-vp 4658 4693 4502 95.93 96.65 96.29 e-sbar 535 535 475 88.79 88.79 88.79 o 6180 6175 5964 96.58 96.50 96.54 e-adjp 438 406 345 84.98 78.77 81.75 i-advp 89 85 57 67.06 64.04 65.52 e-advp 866 838 707 84.37 81.64 82.98 i-adjp 167 122 105 86.07 62.87 72.66 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.60 74.13 75.82 Avg2. 47375 47375 45601 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12426 11697 94.13 94.16 94.15 pp 4811 4839 4703 97.19 97.76 97.47 vp 4658 4693 4386 93.46 94.16 93.81 sbar 535 535 461 86.17 86.17 86.17 adjp 438 406 327 80.54 74.66 77.49 advp 866 838 696 83.05 80.37 81.69 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.02 70.10 73.40 Avg2. 23852 23852 22359 93.74 93.74 93.74 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 112 Log-likelihood = -7428.623346 Norm (log-likelihood gradient vector) = 182.984355 Norm (lambda vector) = 290.953308 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12426 12102 97.39 97.42 97.41 e-pp 4811 4839 4715 97.44 98.00 97.72 i-np 14376 14410 13960 96.88 97.11 96.99 i-vp 2646 2650 2527 95.36 95.50 95.43 e-vp 4658 4694 4503 95.93 96.67 96.30 e-sbar 535 534 476 89.14 88.97 89.06 o 6180 6185 5966 96.46 96.54 96.50 e-adjp 438 406 345 84.98 78.77 81.75 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 840 708 84.29 81.76 83.00 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 108 85 78.70 80.19 79.44 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.63 74.19 75.87 Avg2. 47375 47375 45597 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12426 11690 94.08 94.11 94.09 pp 4811 4839 4704 97.21 97.78 97.49 vp 4658 4694 4387 93.46 94.18 93.82 sbar 535 534 462 86.52 86.36 86.44 adjp 438 406 327 80.54 74.66 77.49 advp 866 840 697 82.98 80.48 81.71 prt 106 108 85 78.70 80.19 79.44 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.06 70.22 73.48 Avg2. 23852 23855 22357 93.72 93.73 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 113 Log-likelihood = -7320.045877 Norm (log-likelihood gradient vector) = 220.608343 Norm (lambda vector) = 292.592069 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12383 12082 97.57 97.26 97.42 e-pp 4811 4860 4724 97.20 98.19 97.69 i-np 14376 14525 14016 96.50 97.50 96.99 i-vp 2646 2651 2526 95.28 95.46 95.37 e-vp 4658 4683 4496 96.01 96.52 96.26 e-sbar 535 516 470 91.09 87.85 89.44 o 6180 6132 5943 96.92 96.17 96.54 e-adjp 438 402 345 85.82 78.77 82.14 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 836 704 84.21 81.29 82.73 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 38 32 84.21 66.67 74.42 e-prt 106 106 83 78.30 78.30 78.30 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.99 73.99 75.94 Avg2. 47375 47375 45599 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12383 11666 94.21 93.91 94.06 pp 4811 4860 4714 97.00 97.98 97.49 vp 4658 4683 4381 93.55 94.05 93.80 sbar 535 516 456 88.37 85.23 86.77 adjp 438 402 327 81.34 74.66 77.86 advp 866 836 693 82.89 80.02 81.43 prt 106 106 83 78.30 78.30 78.30 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.28 69.86 73.38 Avg2. 23852 23794 22325 93.83 93.60 93.71 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 114 Log-likelihood = -7271.295236 Norm (log-likelihood gradient vector) = 772.311981 Norm (lambda vector) = 296.265094 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12407 12093 97.47 97.35 97.41 e-pp 4811 4851 4718 97.26 98.07 97.66 i-np 14376 14456 13983 96.73 97.27 97.00 i-vp 2646 2648 2526 95.39 95.46 95.43 e-vp 4658 4689 4500 95.97 96.61 96.29 e-sbar 535 523 469 89.67 87.66 88.66 o 6180 6168 5958 96.60 96.41 96.50 e-adjp 438 406 346 85.22 79.00 81.99 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 837 707 84.47 81.64 83.03 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 108 85 78.70 80.19 79.44 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.79 74.12 75.91 Avg2. 47375 47375 45595 96.24 96.24 96.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12407 11680 94.14 94.03 94.08 pp 4811 4851 4708 97.05 97.86 97.45 vp 4658 4689 4386 93.54 94.16 93.85 sbar 535 523 455 87.00 85.05 86.01 adjp 438 406 328 80.79 74.89 77.73 advp 866 837 696 83.15 80.37 81.74 prt 106 108 85 78.70 80.19 79.44 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.15 70.10 73.46 Avg2. 23852 23829 22343 93.76 93.67 93.72 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 115 Log-likelihood = -7233.236757 Norm (log-likelihood gradient vector) = 336.988298 Norm (lambda vector) = 294.236662 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12419 12099 97.42 97.40 97.41 e-pp 4811 4850 4719 97.30 98.09 97.69 i-np 14376 14422 13967 96.85 97.15 97.00 i-vp 2646 2653 2529 95.33 95.58 95.45 e-vp 4658 4692 4503 95.97 96.67 96.32 e-sbar 535 527 474 89.94 88.60 89.27 o 6180 6178 5963 96.52 96.49 96.50 e-adjp 438 405 345 85.19 78.77 81.85 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 840 706 84.05 81.52 82.77 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.77 74.12 75.90 Avg2. 47375 47375 45599 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12419 11689 94.12 94.10 94.11 pp 4811 4850 4709 97.09 97.88 97.48 vp 4658 4692 4388 93.52 94.20 93.86 sbar 535 527 460 87.29 85.98 86.63 adjp 438 405 327 80.74 74.66 77.58 advp 866 840 695 82.74 80.25 81.48 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.11 70.08 73.43 Avg2. 23852 23848 22357 93.75 93.73 93.74 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 116 Log-likelihood = -7175.956425 Norm (log-likelihood gradient vector) = 172.700663 Norm (lambda vector) = 295.380600 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12422 12103 97.43 97.43 97.43 e-pp 4811 4852 4720 97.28 98.11 97.69 i-np 14376 14408 13964 96.92 97.13 97.03 i-vp 2646 2653 2529 95.33 95.58 95.45 e-vp 4658 4693 4504 95.97 96.69 96.33 e-sbar 535 526 473 89.92 88.41 89.16 o 6180 6183 5967 96.51 96.55 96.53 e-adjp 438 406 345 84.98 78.77 81.75 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 843 707 83.87 81.64 82.74 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 107 84 78.50 79.25 78.87 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 11 8 72.73 61.54 66.67 e-conjp 9 7 4 57.14 44.44 50.00 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.70 74.12 75.87 Avg2. 47375 47375 45606 96.27 96.27 96.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12422 11696 94.16 94.16 94.16 pp 4811 4852 4709 97.05 97.88 97.46 vp 4658 4693 4388 93.50 94.20 93.85 sbar 535 526 460 87.45 85.98 86.71 adjp 438 406 327 80.54 74.66 77.49 advp 866 843 696 82.56 80.37 81.45 prt 106 107 84 78.50 79.25 78.87 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 7 4 57.14 44.44 50.00 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.09 70.09 73.43 Avg2. 23852 23857 22365 93.75 93.77 93.76 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 117 Log-likelihood = -7153.505854 Norm (log-likelihood gradient vector) = 153.655744 Norm (lambda vector) = 295.647330 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12435 12104 97.34 97.44 97.39 e-pp 4811 4856 4722 97.24 98.15 97.69 i-np 14376 14385 13947 96.96 97.02 96.99 i-vp 2646 2652 2529 95.36 95.58 95.47 e-vp 4658 4694 4504 95.95 96.69 96.32 e-sbar 535 524 472 90.08 88.22 89.14 o 6180 6189 5968 96.43 96.57 96.50 e-adjp 438 407 347 85.26 79.22 82.13 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 842 709 84.20 81.87 83.02 i-adjp 167 123 105 85.37 62.87 72.41 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 40 32 80.00 66.67 72.73 e-prt 106 107 85 79.44 80.19 79.81 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.36 75.67 76.99 Avg2. 47375 47375 45600 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12435 11690 94.01 94.11 94.06 pp 4811 4856 4711 97.01 97.92 97.47 vp 4658 4694 4388 93.48 94.20 93.84 sbar 535 524 459 87.60 85.79 86.69 adjp 438 407 328 80.59 74.89 77.63 advp 866 842 698 82.90 80.60 81.73 prt 106 107 85 79.44 80.19 79.81 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.75 71.33 74.40 Avg2. 23852 23874 22365 93.68 93.77 93.72 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 118 Log-likelihood = -7110.032301 Norm (log-likelihood gradient vector) = 256.986880 Norm (lambda vector) = 296.589188 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12428 12101 97.37 97.42 97.39 e-pp 4811 4848 4719 97.34 98.09 97.71 i-np 14376 14398 13953 96.91 97.06 96.98 i-vp 2646 2653 2530 95.36 95.62 95.49 e-vp 4658 4696 4504 95.91 96.69 96.30 e-sbar 535 528 473 89.58 88.41 88.99 o 6180 6181 5966 96.52 96.54 96.53 e-adjp 438 408 347 85.05 79.22 82.03 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 839 707 84.27 81.64 82.93 i-adjp 167 125 106 84.80 63.47 72.60 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 111 86 77.48 81.13 79.26 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.30 75.75 77.01 Avg2. 47375 47375 45600 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12428 11689 94.05 94.10 94.08 pp 4811 4848 4708 97.11 97.86 97.48 vp 4658 4696 4389 93.46 94.22 93.84 sbar 535 528 460 87.12 85.98 86.55 adjp 438 408 329 80.64 75.11 77.78 advp 866 839 696 82.96 80.37 81.64 prt 106 111 86 77.48 81.13 79.26 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.53 71.43 74.36 Avg2. 23852 23867 22363 93.70 93.76 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 119 Log-likelihood = -7052.366137 Norm (log-likelihood gradient vector) = 169.019896 Norm (lambda vector) = 297.994640 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12429 12101 97.36 97.42 97.39 e-pp 4811 4848 4719 97.34 98.09 97.71 i-np 14376 14397 13954 96.92 97.06 96.99 i-vp 2646 2652 2529 95.36 95.58 95.47 e-vp 4658 4696 4502 95.87 96.65 96.26 e-sbar 535 528 474 89.77 88.60 89.18 o 6180 6182 5966 96.51 96.54 96.52 e-adjp 438 410 348 84.88 79.45 82.08 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 836 705 84.33 81.41 82.84 i-adjp 167 126 106 84.13 63.47 72.35 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 111 86 77.48 81.13 79.26 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.27 75.75 76.99 Avg2. 47375 47375 45598 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12429 11686 94.02 94.08 94.05 pp 4811 4848 4708 97.11 97.86 97.48 vp 4658 4696 4387 93.42 94.18 93.80 sbar 535 528 461 87.31 86.17 86.74 adjp 438 410 329 80.24 75.11 77.59 advp 866 836 694 83.01 80.14 81.55 prt 106 111 86 77.48 81.13 79.26 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.51 71.42 74.34 Avg2. 23852 23867 22357 93.67 93.73 93.70 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 120 Log-likelihood = -7004.200634 Norm (log-likelihood gradient vector) = 152.169192 Norm (lambda vector) = 298.860279 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12421 12100 97.42 97.41 97.41 e-pp 4811 4845 4713 97.28 97.96 97.62 i-np 14376 14433 13974 96.82 97.20 97.01 i-vp 2646 2651 2526 95.28 95.46 95.37 e-vp 4658 4690 4500 95.95 96.61 96.28 e-sbar 535 534 475 88.95 88.79 88.87 o 6180 6167 5960 96.64 96.44 96.54 e-adjp 438 409 349 85.33 79.68 82.41 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 826 699 84.62 80.72 82.62 i-adjp 167 126 106 84.13 63.47 72.35 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 112 86 76.79 81.13 78.90 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.17 75.73 76.93 Avg2. 47375 47375 45596 96.24 96.24 96.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12421 11686 94.08 94.08 94.08 pp 4811 4845 4702 97.05 97.73 97.39 vp 4658 4690 4382 93.43 94.07 93.75 sbar 535 534 462 86.52 86.36 86.44 adjp 438 409 330 80.68 75.34 77.92 advp 866 826 688 83.29 79.45 81.32 prt 106 112 86 76.79 81.13 78.90 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 71.37 74.28 Avg2. 23852 23846 22342 93.69 93.67 93.68 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 121 Log-likelihood = -6919.353347 Norm (log-likelihood gradient vector) = 420.180794 Norm (lambda vector) = 301.425712 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12429 12104 97.39 97.44 97.41 e-pp 4811 4851 4717 97.24 98.05 97.64 i-np 14376 14410 13964 96.90 97.13 97.02 i-vp 2646 2647 2524 95.35 95.39 95.37 e-vp 4658 4693 4501 95.91 96.63 96.27 e-sbar 535 529 473 89.41 88.41 88.91 o 6180 6180 5965 96.52 96.52 96.52 e-adjp 438 409 349 85.33 79.68 82.41 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 829 701 84.56 80.95 82.71 i-adjp 167 126 106 84.13 63.47 72.35 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 111 85 76.58 80.19 78.34 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.17 75.67 76.90 Avg2. 47375 47375 45597 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12429 11694 94.09 94.14 94.11 pp 4811 4851 4706 97.01 97.82 97.41 vp 4658 4693 4383 93.39 94.10 93.74 sbar 535 529 460 86.96 85.98 86.47 adjp 438 409 330 80.68 75.34 77.92 advp 866 829 690 83.23 79.68 81.42 prt 106 111 85 76.58 80.19 78.34 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.44 71.28 74.23 Avg2. 23852 23860 22354 93.69 93.72 93.70 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 122 Log-likelihood = -6851.235587 Norm (log-likelihood gradient vector) = 174.337867 Norm (lambda vector) = 302.251858 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12433 12103 97.35 97.43 97.39 e-pp 4811 4858 4722 97.20 98.15 97.67 i-np 14376 14405 13960 96.91 97.11 97.01 i-vp 2646 2646 2524 95.39 95.39 95.39 e-vp 4658 4693 4500 95.89 96.61 96.25 e-sbar 535 524 473 90.27 88.41 89.33 o 6180 6180 5967 96.55 96.55 96.55 e-adjp 438 410 349 85.12 79.68 82.31 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 831 701 84.36 80.95 82.62 i-adjp 167 125 105 84.00 62.87 71.92 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 110 84 76.36 79.25 77.78 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.24 75.59 76.89 Avg2. 47375 47375 45596 96.24 96.24 96.24 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12433 11690 94.02 94.11 94.07 pp 4811 4858 4711 96.97 97.92 97.45 vp 4658 4693 4382 93.37 94.07 93.72 sbar 535 524 460 87.79 85.98 86.87 adjp 438 410 329 80.24 75.11 77.59 advp 866 831 690 83.03 79.68 81.32 prt 106 110 84 76.36 79.25 77.78 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 71.17 74.17 Avg2. 23852 23868 22352 93.65 93.71 93.68 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 123 Log-likelihood = -6815.021523 Norm (log-likelihood gradient vector) = 135.922771 Norm (lambda vector) = 302.825797 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12432 12103 97.35 97.43 97.39 e-pp 4811 4860 4727 97.26 98.25 97.76 i-np 14376 14406 13963 96.92 97.13 97.03 i-vp 2646 2650 2528 95.40 95.54 95.47 e-vp 4658 4696 4503 95.89 96.67 96.28 e-sbar 535 518 471 90.93 88.04 89.46 o 6180 6178 5966 96.57 96.54 96.55 e-adjp 438 406 347 85.47 79.22 82.23 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 838 704 84.01 81.29 82.63 i-adjp 167 122 105 86.07 62.87 72.66 i-sbar 4 16 3 18.75 75.00 30.00 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 109 83 76.15 78.30 77.21 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.38 75.53 76.93 Avg2. 47375 47375 45608 96.27 96.27 96.27 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12432 11695 94.07 94.15 94.11 pp 4811 4860 4715 97.02 98.00 97.51 vp 4658 4696 4385 93.38 94.14 93.76 sbar 535 518 458 88.42 85.61 86.99 adjp 438 406 328 80.79 74.89 77.73 advp 866 838 693 82.70 80.02 81.34 prt 106 109 83 76.15 78.30 77.21 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.50 71.07 74.14 Avg2. 23852 23868 22363 93.69 93.76 93.73 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 124 Log-likelihood = -6767.636039 Norm (log-likelihood gradient vector) = 168.933523 Norm (lambda vector) = 304.073748 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12474 12120 97.16 97.57 97.37 e-pp 4811 4841 4713 97.36 97.96 97.66 i-np 14376 14322 13917 97.17 96.81 96.99 i-vp 2646 2648 2525 95.35 95.43 95.39 e-vp 4658 4704 4504 95.75 96.69 96.22 e-sbar 535 534 475 88.95 88.79 88.87 o 6180 6213 5980 96.25 96.76 96.51 e-adjp 438 407 347 85.26 79.22 82.13 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 838 702 83.77 81.06 82.39 i-adjp 167 121 104 85.95 62.28 72.22 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 111 84 75.68 79.25 77.42 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.09 75.56 76.80 Avg2. 47375 47375 45579 96.21 96.21 96.21 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12474 11708 93.86 94.25 94.06 pp 4811 4841 4701 97.11 97.71 97.41 vp 4658 4704 4385 93.22 94.14 93.68 sbar 535 534 462 86.52 86.36 86.44 adjp 438 407 327 80.34 74.66 77.40 advp 866 838 691 82.46 79.79 81.10 prt 106 111 84 75.68 79.25 77.42 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.17 71.17 74.05 Avg2. 23852 23918 22364 93.50 93.76 93.63 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 125 Log-likelihood = -6791.083649 Norm (log-likelihood gradient vector) = 578.545687 Norm (lambda vector) = 306.519553 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12443 12109 97.32 97.48 97.40 e-pp 4811 4853 4722 97.30 98.15 97.72 i-np 14376 14387 13953 96.98 97.06 97.02 i-vp 2646 2647 2525 95.39 95.43 95.41 e-vp 4658 4698 4504 95.87 96.69 96.28 e-sbar 535 524 472 90.08 88.22 89.14 o 6180 6186 5969 96.49 96.59 96.54 e-adjp 438 407 347 85.26 79.22 82.13 i-advp 89 83 57 68.67 64.04 66.28 e-advp 866 838 703 83.89 81.18 82.51 i-adjp 167 121 104 85.95 62.28 72.22 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 110 84 76.36 79.25 77.78 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.26 75.54 76.88 Avg2. 47375 47375 45600 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12443 11700 94.03 94.19 94.11 pp 4811 4853 4710 97.05 97.90 97.48 vp 4658 4698 4385 93.34 94.14 93.74 sbar 535 524 459 87.60 85.79 86.69 adjp 438 407 327 80.34 74.66 77.40 advp 866 838 692 82.58 79.91 81.22 prt 106 110 84 76.36 79.25 77.78 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.38 71.14 74.13 Avg2. 23852 23882 22363 93.64 93.76 93.70 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 126 Log-likelihood = -6752.521778 Norm (log-likelihood gradient vector) = 232.080515 Norm (lambda vector) = 304.909975 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12440 12109 97.34 97.48 97.41 e-pp 4811 4848 4718 97.32 98.07 97.69 i-np 14376 14389 13958 97.00 97.09 97.05 i-vp 2646 2649 2526 95.36 95.46 95.41 e-vp 4658 4702 4505 95.81 96.72 96.26 e-sbar 535 527 473 89.75 88.41 89.08 o 6180 6183 5969 96.54 96.59 96.56 e-adjp 438 405 346 85.43 79.00 82.09 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 838 702 83.77 81.06 82.39 i-adjp 167 121 104 85.95 62.28 72.22 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 111 84 75.68 79.25 77.42 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.16 75.53 76.83 Avg2. 47375 47375 45602 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12440 11705 94.09 94.23 94.16 pp 4811 4848 4706 97.07 97.82 97.44 vp 4658 4702 4386 93.28 94.16 93.72 sbar 535 527 460 87.29 85.98 86.63 adjp 438 405 326 80.49 74.43 77.34 advp 866 838 691 82.46 79.79 81.10 prt 106 111 84 75.68 79.25 77.42 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.29 71.12 74.08 Avg2. 23852 23880 22364 93.65 93.76 93.71 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 127 Log-likelihood = -6703.578836 Norm (log-likelihood gradient vector) = 221.456279 Norm (lambda vector) = 306.634043 Log-likelihood and gradient computational time: 78 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12438 12108 97.35 97.47 97.41 e-pp 4811 4853 4720 97.26 98.11 97.68 i-np 14376 14395 13960 96.98 97.11 97.04 i-vp 2646 2650 2526 95.32 95.46 95.39 e-vp 4658 4704 4506 95.79 96.74 96.26 e-sbar 535 524 472 90.08 88.22 89.14 o 6180 6178 5966 96.57 96.54 96.55 e-adjp 438 404 346 85.64 79.00 82.19 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 835 699 83.71 80.72 82.19 i-adjp 167 120 104 86.67 62.28 72.47 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 112 85 75.89 80.19 77.98 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.24 75.56 76.87 Avg2. 47375 47375 45600 96.25 96.25 96.25 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12438 11702 94.08 94.20 94.14 pp 4811 4853 4708 97.01 97.86 97.43 vp 4658 4704 4386 93.24 94.16 93.70 sbar 535 524 459 87.60 85.79 86.69 adjp 438 404 326 80.69 74.43 77.43 advp 866 835 688 82.40 79.45 80.89 prt 106 112 85 75.89 80.19 77.98 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.34 71.16 74.12 Avg2. 23852 23879 22360 93.64 93.74 93.69 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 99 seconds Iteration: 128 Log-likelihood = -6589.015887 Norm (log-likelihood gradient vector) = 180.774435 Norm (lambda vector) = 309.612991 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12410 12096 97.47 97.38 97.42 e-pp 4811 4847 4717 97.32 98.05 97.68 i-np 14376 14445 13987 96.83 97.29 97.06 i-vp 2646 2651 2526 95.28 95.46 95.37 e-vp 4658 4703 4503 95.75 96.67 96.21 e-sbar 535 525 471 89.71 88.04 88.87 o 6180 6160 5959 96.74 96.42 96.58 e-adjp 438 406 347 85.47 79.22 82.23 i-advp 89 85 57 67.06 64.04 65.52 e-advp 866 829 697 84.08 80.48 82.24 i-adjp 167 124 106 85.48 63.47 72.85 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 112 86 76.79 81.13 78.90 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.17 75.66 76.89 Avg2. 47375 47375 45603 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12410 11694 94.23 94.14 94.18 pp 4811 4847 4705 97.07 97.80 97.43 vp 4658 4703 4383 93.20 94.10 93.64 sbar 535 525 458 87.24 85.61 86.42 adjp 438 406 327 80.54 74.66 77.49 advp 866 829 686 82.75 79.21 80.94 prt 106 112 86 76.79 81.13 78.90 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 71.22 74.20 Avg2. 23852 23841 22345 93.73 93.68 93.70 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 129 Log-likelihood = -6549.721638 Norm (log-likelihood gradient vector) = 525.357607 Norm (lambda vector) = 313.533386 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 77 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12427 12105 97.41 97.45 97.43 e-pp 4811 4851 4719 97.28 98.09 97.68 i-np 14376 14417 13971 96.91 97.18 97.04 i-vp 2646 2650 2526 95.32 95.46 95.39 e-vp 4658 4703 4505 95.79 96.72 96.25 e-sbar 535 523 471 90.06 88.04 89.04 o 6180 6171 5963 96.63 96.49 96.56 e-adjp 438 406 347 85.47 79.22 82.23 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 830 697 83.98 80.48 82.19 i-adjp 167 124 106 85.48 63.47 72.85 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 111 85 76.58 80.19 78.34 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.22 75.61 76.89 Avg2. 47375 47375 45603 96.26 96.26 96.26 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12427 11699 94.14 94.18 94.16 pp 4811 4851 4707 97.03 97.84 97.43 vp 4658 4703 4385 93.24 94.14 93.69 sbar 535 523 458 87.57 85.61 86.58 adjp 438 406 327 80.54 74.66 77.49 advp 866 830 686 82.65 79.21 80.90 prt 106 111 85 76.58 80.19 78.34 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 71.14 74.15 Avg2. 23852 23860 22353 93.68 93.72 93.70 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds Iteration: 130 Log-likelihood = -6525.303156 Norm (log-likelihood gradient vector) = 220.351162 Norm (lambda vector) = 311.552303 Log-likelihood and gradient computational time: 77 seconds Training iteration elapsed: 78 seconds Label-based performance evaluation: Label Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- e-np 12422 12429 12107 97.41 97.46 97.44 e-pp 4811 4850 4718 97.28 98.07 97.67 i-np 14376 14412 13973 96.95 97.20 97.08 i-vp 2646 2650 2526 95.32 95.46 95.39 e-vp 4658 4700 4505 95.85 96.72 96.28 e-sbar 535 525 471 89.71 88.04 88.87 o 6180 6176 5967 96.62 96.55 96.58 e-adjp 438 405 347 85.68 79.22 82.33 i-advp 89 84 57 67.86 64.04 65.90 e-advp 866 831 697 83.87 80.48 82.14 i-adjp 167 123 106 86.18 63.47 73.10 i-sbar 4 17 3 17.65 75.00 28.57 i-pp 48 39 32 82.05 66.67 73.56 e-prt 106 112 86 76.79 81.13 78.90 e-lst 5 0 0 0.00 0.00 0.00 i-intj 0 0 0 0.00 0.00 0.00 e-intj 2 1 1 100.00 50.00 66.67 i-conjp 13 13 10 76.92 76.92 76.92 e-conjp 9 8 5 62.50 55.56 58.82 i-prt 0 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 ----- ------ ----- ----- ------- ------- ------------- Avg1. 78.26 75.67 76.94 Avg2. 47375 47375 45611 96.28 96.28 96.28 Chunk-based performance evaluation: Chunk Manual Model Match Pre.(%) Rec.(%) F1-Measure(%) ----- ------ ----- ----- ------- ------- ------------- np 12422 12429 11708 94.20 94.25 94.23 pp 4811 4850 4706 97.03 97.82 97.42 vp 4658 4700 4385 93.30 94.14 93.72 sbar 535 525 458 87.24 85.61 86.42 adjp 438 405 327 80.74 74.66 77.58 advp 866 831 686 82.55 79.21 80.85 prt 106 112 86 76.79 81.13 78.90 lst 5 0 0 0.00 0.00 0.00 intj 2 1 1 100.00 50.00 66.67 conjp 9 8 5 62.50 55.56 58.82 ucp 0 0 0 0.00 0.00 0.00 ----- ------ ----- ----- ------- ------- ------------- Avg1. 77.43 71.24 74.21 Avg2. 23852 23861 22362 93.72 93.75 93.74 Current max chunk-based F1: 94.05 (iteration 63) Training iteration elapsed (including evaluation time): 98 seconds The training process elapsed: 12752 seconds