Aspect Sentiment Triplet Extraction on Rest SemEval 2015 (test)
68.03F1 ScoreMvP + FiVeD-unified
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MvP + FiVeD-unifiedVerification Model=DeepSeek-Reasoner, Paradigm=Generative2026.05 | 68.03 | 66.95 | 69.24 | |
| MvP + FiVeD-unifiedVerification Model=Qwen, Paradigm=Generative2026.05 | 67.82 | 66.49 | 69.28 | |
| TAGSParadigm=Sequence tagging2026.05 | 67.33 | 69.55 | 65.25 | |
| MvP + FiVeDVerification Model=Qwen, Paradigm=Generative2026.05 | 66.96 | 66.72 | 67.26 | |
| MvP + FiVeDVerification Model=DeepSeek-Reasoner, Paradigm=Generative2026.05 | 66.87 | 66.59 | 67.26 | |
| Paraphrase + FiVeD-unifiedVerification Model=Qwen, Paradigm=Generative2026.05 | 66.06 | 64.05 | 68.29 | |
| D2E2SParadigm=Span-based2026.05 | 65.86 | 70.09 | 62.11 | |
| Paraphrase + FiVeD-unifiedVerification Model=DeepSeek-Reasoner, Paradigm=Generative2026.05 | 65.82 | 64.37 | 67.46 | |
| GAS + FiVeD-unifiedVerification Model=DeepSeek-Reasoner, Paradigm=Generative2026.05 | 65.39 | 59.74 | 72.37 | |
| Paraphrase + FiVeDVerification Model=DeepSeek-Reasoner, Paradigm=Generative2026.05 | 65.36 | 65.11 | 65.65 | |
| Paraphrase + FiVeDVerification Model=Qwen, Paradigm=Generative2026.05 | 65.31 | 64.11 | 66.64 | |
| MiniConGTSParadigm=Table tagging2026.05 | 65.15 | 66.5 | 63.86 | |
| GAS + FiVeD-unifiedVerification Model=Qwen, Paradigm=Generative2026.05 | 65.13 | 60.98 | 69.98 | |
| SimSTARParadigm=Table tagging2026.05 | 65.09 | 71.71 | 59.59 | |
| MvPParadigm=Generative2026.05 | 64.99 | 64.5 | 65.48 | |
| COM-MRCParadigm=MRC-based2026.05 | 64.53 | 68.35 | 61.24 | |
| LEGOParadigm=Generative2026.05 | 64.4 | — | — | |
| Span-ASTEEncoder=BERT2021.07 | 63.27 | 62.18 | 64.45 | |
| Span-ASTEParadigm=Span-based2026.05 | 63.27 | 62.18 | 64.45 | |
| GAS + FiVeDVerification Model=Qwen, Paradigm=Generative2026.05 | 63.05 | 57.56 | 69.77 | |
| GAS + FiVeDVerification Model=DeepSeek-Reasoner, Paradigm=Generative2026.05 | 63.01 | 58.15 | 68.91 | |
| Triple-MRCParadigm=MRC-based2026.05 | 62.86 | — | — | |
| PARAPHRASEType=Sequence-to-sequence, Modeling=Natural Language Labels2021.10 | 62.56 | — | — | |
| ParaphraseParadigm=Generative2026.05 | 61.85 | 65.48 | 57.23 | |
| GASType=Sequence-to-sequence2021.10 | 60.23 | — | — | |
| GTSEncoder=BERT2021.07 | 60.15 | 62.59 | 57.94 | |
| GASParadigm=Generative2026.05 | 60.05 | 57.48 | 62.97 | |
| Span-BARTParadigm=Sequence tagging2026.05 | 59.26 | 59.14 | 59.38 | |
| EMC-GCNParadigm=Table tagging2026.05 | 58.81 | 61.7 | 56.26 | |
| PBF -w/o A2021.03 | 58.8 | 56.4 | 61.6 | |
| Two-StageType=Two-stage2021.10 | 58.59 | — | — | |
| PBF2021.03 | 58.5 | 55.8 | 61.5 | |
| PBF-M12021.03 | 58.4 | 57.8 | 59.4 | |
| PBF -w/o P2021.03 | 58.2 | 56.2 | 60.4 | |
| PBF-M32021.03 | 57.7 | 54.2 | 61.7 | |
| Span-ASTEEncoder=BiLSTM2021.07 | 57.56 | 64.29 | 52.12 | |
| JETEncoder=BERT2021.07 | 57.53 | 64.45 | 51.96 | |
| Jet+BERTType=End-to-end, Backbone=BERT2021.10 | 57.53 | — | — | |
| Dual-MRCParadigm=MRC-based2026.05 | 57.21 | 63.78 | 51.87 | |
| GTS-BERT2021.03 | 56.4 | 61.8 | 52 | |
| GTS+BERTType=End-to-end, Backbone=BERT2021.10 | 54.88 | — | — | |
| JET-bert-o2021.03 | 53.3 | 53.8 | 52.9 | |
| PBF-M22021.03 | 53 | 50.2 | 56.2 | |
| GTSEncoder=BiLSTM2021.07 | 52.66 | 60.1 | 46.89 | |
| JETEncoder=BiLSTM2021.07 | 52.5 | 64.37 | 44.33 | |
| GTS-BiLSTM2021.03 | 52.4 | 67.3 | 42.9 | |
| Peng et al.Encoder=BiLSTM2021.07 | 52.32 | 48.07 | 57.51 | |
| P-pipelineType=Pipeline-based2021.10 | 52.32 | — | — | |
| Peng-two-stageParadigm=Sequence tagging2026.05 | 52.32 | 48.07 | 57.51 | |
| GPT-4oEvaluation Prompting Strategy=few-shot, Paradigm=Generative2026.05 | 51.95 | 45.57 | 60.41 | |
| JETo2021.03 | 51.8 | 57.5 | 47.2 | |
| GTS-CNN2021.03 | 51.8 | 56.3 | 48.1 | |
| GPT-3.5-turboEvaluation Prompting Strategy=few-shot, Paradigm=Generative2026.05 | 51.39 | 43.34 | 63.09 | |
| JET-bert-t2021.03 | 49.4 | 50.7 | 48.2 | |
| OTE-MTL2021.03 | 49.3 | 56.3 | 44 | |
| Zhang et al.Encoder=BiLSTM2021.07 | 47.94 | 55.63 | 42.51 | |
| Li-unified-REncoder=BiLSTM2021.07 | 47.82 | 44.72 | 51.39 | |
| Li-unified-RType=Pipeline-based2021.10 | 47.82 | — | — | |
| GPT-3.5-turboEvaluation Prompting Strategy=CoT, Paradigm=Generative2026.05 | 46.57 | 39.51 | 56.7 | |
| JETt2021.03 | 46.5 | 45.3 | 47.8 | |
| GPT-3.5-turboEvaluation Prompting Strategy=CoT + few-shot, Paradigm=Generative2026.05 | 46.02 | 39.02 | 56.08 | |
| PBF -w/o AP2021.03 | 45.7 | 41.7 | 50.9 | |
| GPT-3.5-turboEvaluation Prompting Strategy=zero-shot, Paradigm=Generative2026.05 | 43.02 | 36.02 | 53.4 | |
| GPT-4oEvaluation Prompting Strategy=CoT + few-shot, Paradigm=Generative2026.05 | 42.47 | 35.08 | 53.81 | |
| GPT-4oEvaluation Prompting Strategy=CoT, Paradigm=Generative2026.05 | 40.1 | 33.07 | 50.93 | |
| CMLA+Encoder=BiLSTM2021.07 | 37.01 | 34.56 | 39.84 | |
| CMLA+Type=Pipeline-based2021.10 | 37.01 | — | — | |
| GPT-4oEvaluation Prompting Strategy=zero-shot, Paradigm=Generative2026.05 | 32.05 | 27.85 | 37.73 | |
| RINANTE+Encoder=BiLSTM2021.07 | 29.97 | 29.88 | 30.06 |