Table 13 The ablation experiment results on four datasets - F1 score
From: Chinese medical entity recognition based on the dual-branch TENER model
Model | CCKS2019 F1 score (%) | ERTCMM F1 score (%) | CMeEE F1 score (%) | CEMR F1 score (%) |
|---|---|---|---|---|
Ours | 86.41 | 74.37 | 68.39 | 70.16 |
- Entity dictionary | 82.13 | 70.26 | 63.55 | 67.29 |
TENER\(\longrightarrow\)BERT | 80.35 | 64,93 | 61.04 | 63.05 |
Single | 85.37 | 68.32 | 64.96 | 68.21 |
- Radicals feature | 83.75 | 71.02 | 64.47 | 66.19 |
TENER\(\longrightarrow\)Glove 100d | 84.38 | 66.32 | 63.96 | 67.36 |
A-Softmax\(\longrightarrow\)Softmax | 86.02 | 69.57 | 65.75 | 69.23 |
- Semi-supervised | 86.33 | 73.41 | 65.83 | 70.03 |