Table 3 The effectiveness of multi level attention for event extraction
From: A biomedical event extraction method based on fine-grained and attention mechanism
Model | Precision (%) | Recall (%) | F-score (%) |
|---|---|---|---|
Bi-LSTM | 90.93 ± 0.35 | 38.50 ± 0.41 | 54.09 ± 0.37 |
Bi-LSTM + WAtt | 90.75 ± 0.22 | 43.00 ± 0.25 | 58.35 ± 0.23 |
Bi-LSTM + SAtt | 89.69 ± 0.23 | 44.12 ± 0.28 | 59.14 ± 0.27 |
Bi-LSTM + MultiAtt | 90.24 ± 0.19 | 44.50 ± 0.16 | 59.61 ± 0.18 |
Bi-LSTM + MultiAtt + Fine-grained | 91.05 ± 0.27 | 44.68 ± 0.31 | 59.94 ± 0.29 |