Table 5.
Ratio of sentences classified by the top 3 classifiers as ci x or ci y that were classified correctly.
| Label and classifier | c1 (%) | c2 (%) | c3 (%) | c4 (%) | c5 (%) | c6 (%) | c7 (%) | c8 (%) | |
| c1x | |||||||||
|
|
CNN-Wa | 34.78% | 60.19% b | 0.00% | 62.50% | 50.26% | 66.81% | 57.14% | 65.91% |
|
|
Matsumoto | 46.67% | 43.40% | 50.00% | 66.23% | 55.31% | 71.10% | 67.27% | 77.27% |
|
|
DTCcCNN-W | 16.04% | 41.66% | 10.00% | 20.69% | 22.58% | 43.59% | 23.73% | 21.52% |
| c1y | |||||||||
|
|
CNN-W | 40.00% | 41.52% | 50.56% | 22.83% | 28.70% | 41.27% | 29.41% | 59.06% |
|
|
Matsumoto | 58.82% | 34.18% | 56.52% | 34.62% | 25.64% | 49.53% | 53.33% | 70.99% |
|
|
DTCCNN-W | 10.98% | 13.50% | 28.57% | 13.38% | 14.25% | 22.90% | 14.29% | 29.96% |
aCNN-W: Convolutional Neural Network with Word2Vec.
bFor each ci , the highest value for both ci x and ci y are italicized for emphasis.
cDTC: dependency tree classifier.