Table 8 Comparison of F1-score for different models on various relation categories
From: A hybrid method based on semi-supervised learning for relation extraction in Chinese EMRs
Models | DAP | SAP | SNAP | TeRD | TeAP | TeCP | TeRS | TeAS | TrAD | TrRD | TrAP |
|---|---|---|---|---|---|---|---|---|---|---|---|
CNN | 64.86 | 78.04 | 75.95 | 79.51 | 85 | 74.29 | 80.4 | 66.76 | 83.9 | 72.74 | 82.36 |
CNN-Att | 70.33 | 82.89 | 84.79 | 77.16 | 84.98 | 80.1 | 84.89 | 75.67 | 83.73 | 83.71 | 83.42 |
PCNN | 69.53 | 84.74 | 80.06 | 75.69 | 82.28 | 76.36 | 79.95 | 75.43 | 84.74 | 83.44 | 83.05 |
ResNet | 75.98 | 91.11 | 86.95 | 84.18 | 89.89 | 78.59 | 85.86 | 86 | 94.99 | 87.62 | 86.24 |
BiLSTM-Att | 74.96 | 92.83 | 92.48 | 84.84 | 85.27 | 73.57 | 85.6 | 87.47 | 92.48 | 80.34 | 87.43 |
BiGRU-Att | 75.34 | 90.62 | 90 | 87.43 | 89.93 | 83.18 | 85.27 | 81.84 | 92.08 | 82.24 | 87.44 |
ResGRU | 78.53 | 92 | 93.37 | 92.05 | 85 | 83.39 | 84.46 | 80.13 | 91.83 | 87.11 | 90.17 |
ResGRU-Att | 80.95 | 93.91 | 92.96 | 88.43 | 86.54 | 85.58 | 87.96 | 94.74 | 93.01 | 87.58 | 95.48 |