Table 6.
Experimental results on HardTest. Overall denotes the macro scores. The highest scores are highlighted in bold.
| Classifier | Hard safe (N = 652) | Hard offensive (N = 663) | Overall (N = 1315) | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| P | R | F1 | P | R | F1 | P | R | F1 | Acc. | |
| COLD Mac | 0.6072 | 0.5169 | 0.5584 | 0.5855 | 0.6712 | 0.6254 | 0.5964 | 0.594 | 0.5919 | 0.5947 |
| COLD-R Mac | 0.5877 | 0.6012 | 0.5944 | 0.5988 | 0.5852 | 0.5919 | 0.5932 | 0.5932 | 0.5932 | 0.5932 |
| CDialBias Mac | 0.5601 | 0.8788 | 0.6842 | 0.7295 | 0.3213 | 0.4461 | 0.6448 | 0.6001 | 0.5651 | 0.5977 |
| TransJigsaw Mac | 0.568 | 0.7561 | 0.6487 | 0.6443 | 0.4344 | 0.5189 | 0.6061 | 0.5953 | 0.5838 | 0.5939 |
| TransSIBC Mac | 0.5694 | 0.4908 | 0.5272 | 0.5591 | 0.635 | 0.5946 | 0.5642 | 0.5629 | 0.5609 | 0.5635 |
| TransCN Mac | 0.5357 | 0.9202 | 0.6772 | 0.7333 | 0.2157 | 0.3333 | 0.6345 | 0.568 | 0.5053 | 0.565 |
| Mix Mac | 0.6252 | 0.6012 | 0.613 | 0.6221 | 0.6456 | 0.6336 | 0.6236 | 0.6234 | 0.6233 | 0.6236 |
| Maj-MultiT | 0.6105 | 0.7117 | 0.6572 | 0.6613 | 0.5535 | 0.6026 | 0.6359 | 0.6326 | 0.6299 | 0.6319 |
| Avg-MultiT | 0.5925 | 0.7807 | 0.6737 | 0.6864 | 0.4721 | 0.5594 | 0.6395 | 0.6264 | 0.6166 | 0.6251 |
| MuDA(γ = 0.0) | 0.565 | 0.8466 | 0.6777 | 0.7041 | 0.359 | 0.359 | 0.6346 | 0.6028 | 0.5766 | 0.6008 |
| MuDA(γ = 0.7) | 0.5847 | 0.773 | 0.6658 | 0.6733 | 0.46 | 0.5466 | 0.629 | 0.6165 | 0.6062 | 0.6152 |
| MuDA(γ = 1.0) | 0.5951 | 0.7776 | 0.6742 | 0.6868 | 0.4796 | 0.5648 | 0.6409 | 0.6286 | 0.6195 | 0.6274 |
| MuDA Mix(γ = 0.5) | 0.6188 | 0.6871 | 0.6512 | 0.6548 | 0.5837 | 0.6172 | 0.6368 | 0.6354 | 0.6342 | 0.6350 |