Table 5 Performance of different models on the long sentence dependency test
From: Biomedical relation extraction method based on ensemble learning and attention mechanism
Model | PPI | DDI | ChemProt | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Â | P | R | F | P | R | F | P | R | F |
BioBERT | 78.2 | 75.0 | 76.6 | 78.5 | 74.2 | 76.3 | 77.6 | 75.1 | 76.3 |
BlueBERT | 76.9 | 72.8 | 74.8 | 77.2 | 73.5 | 75.3 | 75.9 | 71.2 | 73.5 |
PubMedBERT | 79.1 | 76.3 | 77.7 | 78.8 | 76.0 | 77.3 | 78.3 | 75.4 | 76.8 |
BlueBERT(-M) | 75.8 | 73.1 | 74.4 | 77.3 | 74.0 | 75.6 | 74.9 | 71.5 | 73.2 |
BioBERT_SSL_Att | 81.0 | 77.9 | 79.4 | 79.6 | 76.8 | 78.2 | 78.9 | 75.7 | 77.3 |
PubMedBERT_SSL_Att | 80.3 | 77.2 | 78.7 | 79.9 | 77.1 | 78.5 | 79.3 | 76.0 | 77.6 |
BioBERT+CLEK | 77.5 | 73.9 | 75.7 | 78.2 | 74.5 | 76.3 | 78.0 | 75.0 | 76.4 |
PubMedBERT+CLEK | 79.0 | 75.3 | 77.1 | 79.0 | 75.7 | 77.3 | 79.2 | 75.8 | 77.5 |
SARE | 82.3 | 79.1 | 80.7 | 82.0 | 78.9 | 80.4 | 81.1 | 76.9 | 79.0 |