| 模态 | 模型 | TWITTER-2015 | TWITTER-2017 | ||
|---|---|---|---|---|---|
| Accuracy | Macro-F1 | Accuracy | Macro-F1 | ||
| 文本 | BERT-SPC | 73.19 | 64.48 | 63.94 | 61.91 |
| AEN-BERT | 70.97 | 62.27 | 60.05 | 58.12 | |
| TD-BERT | 75.99 | 69.77 | 68.31 | 64.64 | |
| ToiGNN | 74.90 | 68.93 | 67.26 | 64.75 | |
| CL-ABSA | 78.30 | 74.17 | 69.54 | 64.88 | |
| 文本+图像 | MIMN | 71.84 | 65.69 | 65.88 | 62.99 |
| TomBERT | 77.15 | 71.75 | 70.50 | 68.04 | |
| ModalNet-BERT | 79.03 | 72.50 | 72.36 | 69.19 | |
| HIMT | 78.14 | 73.68 | 71.14 | 69.16 | |
| ESAFN | 73.38 | 67.37 | 67.83 | 64.22 | |
| SMP | 77.53 | 72.24 | 71.15 | 69.47 | |
| MCCL-ABSA | 79.85 | 76.24 | 73.18 | 69.72 | |
