Dialogue Aspect-based Sentiment Quadruple Extraction on DiaASQ ZH (test)
48.61Pair Extraction F1 (Target-Aspect)DiaASQ model
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DiaASQ modelPLM=Standard2022.11 | 48.61 | 90.23 | 76.94 | 59.35 | 43.31 | 45.44 | 34.94 | 37.51 | |
| Span-ASTEPLM=Standard2022.11 | 44.13 | — | — | — | 34.46 | 32.21 | 27.42 | 30.85 | |
| SpERTPLM=Standard2022.11 | 38.05 | 90.69 | 76.81 | 54.06 | 31.28 | 21.89 | 13 | 14.19 | |
| ParaPhrasePLM=mT5-base2022.11 | 37.81 | — | — | — | 34.32 | 27.76 | 23.27 | 27.98 | |
| DiaASQ modelPLM=word2vec2022.11 | 34.72 | 85.52 | 75.21 | 47.15 | 26.16 | 30.73 | 14.21 | 17.55 | |
| CRF-Extract-ClassifyPLM=Standard2022.11 | 32.47 | 91.11 | 75.24 | 50.06 | 26.78 | 18.9 | 8.81 | 9.25 | |
| Span-ASTEPLM=word2vec2022.11 | 28.36 | — | — | — | 24.81 | 22.5 | 8.96 | 11.21 |