Table 10 Comparative performance (recall) of different methods for rare entities
From: Multi-task learning for Chinese clinical named entity recognition with external knowledge
Method | 0 | 1 | 2 | 3 |
|---|---|---|---|---|
CCKS2017 | ||||
BERT(Baseline1) | 51.47 | 69.29 | 81.36 | 90.34 |
FT-BERT(Baseline2) | 53.62 | 71.36 | 82.54 | 90.97 |
Our | 53.92 | 76.65 | 85.31 | 91.37 |
FCCd | ||||
BERT(Baseline1) | 50.37 | 59.47 | 71.36 | 84.91 |
FT-BERT(Baseline2) | 52.56 | 60.38 | 73.82 | 84.75 |
Our | 55.97 | 62.74 | 77.47 | 85.11 |