Table 4 Performance comparison of different models
Model | macro-P | macro-R | macro-F1 |
|---|---|---|---|
Bi-LSTM | 83.97% | 83.67% | 83.56% |
CNN_Attention | 84.80% | 84.75% | 84.66% |
2 layers-Bi-LSTM | 84.40% | 84.08% | 83.98% |
2 layers-Bi-LSTM + character embedding | 85.50% | 85.55% | 85.45% |
Model | macro-P | macro-R | macro-F1 |
|---|---|---|---|
Bi-LSTM | 83.97% | 83.67% | 83.56% |
CNN_Attention | 84.80% | 84.75% | 84.66% |
2 layers-Bi-LSTM | 84.40% | 84.08% | 83.98% |
2 layers-Bi-LSTM + character embedding | 85.50% | 85.55% | 85.45% |