Dialogue Aspect-based Sentiment Quadruple Extraction on DiaASQ EN (test)
0.4791Pair F1 (Target-Aspect)DiaASQ model
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DiaASQ modelPLM=Standard2022.11 | 0.4791 | 0.8862 | 0.7471 | 0.6022 | 0.4558 | 0.4427 | 0.3331 | 0.368 | |
| Span-ASTEPLM=Standard2022.11 | 0.4219 | — | — | — | 0.3044 | 0.459 | 0.2699 | 0.2834 | |
| ParaPhrasePLM=mT5-base2022.11 | 0.3722 | — | — | — | 0.3219 | 0.3078 | 0.2454 | 0.2676 | |
| CRF-Extract-ClassifyPLM=Standard2022.11 | 0.3431 | 0.8831 | 0.7171 | 0.479 | 0.2094 | 0.1921 | 0.1159 | 0.128 | |
| DiaASQ modelPLM=word2vec2022.11 | 0.3253 | 0.8302 | 0.6889 | 0.5387 | 0.3109 | 0.3559 | 0.1568 | 0.1957 | |
| SpERTPLM=Standard2022.11 | 0.2833 | 0.8782 | 0.7465 | 0.5417 | 0.2139 | 0.2364 | 0.1307 | 0.1338 | |
| Span-ASTEPLM=word2vec2022.11 | 0.2726 | — | — | — | 0.2063 | 0.4462 | 0.1384 | 0.1417 |