Dialogue Emotion Detection on EmoryNLP
47.61Weighted Avg F1SELF-EMO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SELF-EMOBackbone=Qwen3-8B, External FREE=YES, Training Protocol=Dataset-Unified Setting2026.04 | 47.61 | — | — | 47.87 | |
| SELF-EMOBackbone=Qwen3-4B, External FREE=YES, Training Protocol=Dataset-Unified Setting2026.04 | 44.98 | — | — | 45.52 | |
| TODKATAverage of ten runs=true2021.06 | 43.12 | 42.68 | — | — | |
| Qwen3-8BBackbone=Qwen3-8B, External FREE=YES, Training Protocol=Dataset-Unified Setting2026.04 | 42.71 | — | — | 43.19 | |
| TODKAT (-Topics)Average of ten runs=true, latent topics=false2021.06 | 41.8 | 40.55 | — | — | |
| BiosERCBackbone=LLaMA2-13B, External FREE=NO, Training Protocol=Dataset-Specific Setting2026.04 | 41.68 | — | — | — | |
| Qwen3-4BBackbone=Qwen3-4B, External FREE=YES, Training Protocol=Dataset-Unified Setting2026.04 | 41.5 | — | — | 41.87 | |
| InstructERCBackbone=LLaMA2-7B, External FREE=NO, Training Protocol=Dataset-Specific Setting2026.04 | 41.39 | — | — | — | |
| Llama2Parameters=12.5M(7B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + InstructERC2023.09 | 41.37 | — | — | — | |
| KATCOMETAverage of ten runs=true, knowledge base=COMET2021.06 | 41.1 | 39.74 | — | — | |
| SPCLParameters=356.7M, Extra Knowledge=✗, Model type=transformer, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 40.25 | — | — | — | |
| EACLArchitecture Type=Sequence-based, Backbone=SimCSE-Roberta-Large2024.03 | 40.24 | — | — | — | |
| DAG-ERC + SBase Model=DAG-ERC, Grayscale Label Type=self2022.06 | 40.23 | — | — | — | |
| COSMIC+CKCLArchitecture Type=Sequence-based2024.03 | 40.23 | — | — | — | |
| DIEUArchitecture Type=Sequence-based2024.03 | 40.12 | — | — | — | |
| EmotionICExtra Knowledge=✗, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 40.01 | — | — | — | |
| COMPM + SABase Model=COMPM, Grayscale Label Type=self-adjust2022.06 | 39.95 | — | — | — | |
| SIGATArchitecture Type=Graph-based2024.03 | 39.95 | — | — | — | |
| DAG-ERC + SABase Model=DAG-ERC, Grayscale Label Type=self-adjust2022.06 | 39.85 | — | — | — | |
| DAG-ERC+HCLArchitecture Type=Graph-based2024.03 | 39.82 | — | — | — | |
| HCANParameters=3.5M, Extra Knowledge=✗, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 39.67 | — | — | — | |
| SACLArchitecture Type=Sequence-based2024.03 | 39.65 | — | — | — | |
| SACL-LSTMParameters=2.6M, Extra Knowledge=✗, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 39.65 | — | — | — | |
| SPCL+CLArchitecture Type=Sequence-based, Backbone=SimCSE-Roberta-Large2024.03 | 39.52 | — | — | — | |
| LlamaParameters=12.5M(7B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + InstructERC2023.09 | 39.34 | — | — | — | |
| SupConArchitecture Type=Sequence-based2024.03 | 39.28 | — | — | — | |
| ChatGLM2Parameters=12.5M(6B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + InstructERC2023.09 | 39.09 | — | — | — | |
| COMPM + SBase Model=COMPM, Grayscale Label Type=self2022.06 | 39.05 | — | — | — | |
| CoG-BART2021.12 | 39.04 | 42.58 | — | — | |
| Cog-BARTArchitecture Type=Sequence-based2024.03 | 39.04 | — | — | — | |
| DAG-ERCBase Model=DAG-ERC, Grayscale Label Type=None2022.06 | 39.02 | — | — | — | |
| DAG-ERCArchitecture Type=Graph-based2024.03 | 39.02 | — | — | — | |
| DAG-ERCBackbone=-, External FREE=-, Training Protocol=Dataset-Specific Setting2026.04 | 39.02 | — | — | — | |
| Psychological + SBase Model=Psychological, Grayscale Label Type=self2022.06 | 38.96 | — | — | — | |
| COMPMArchitecture Type=Sequence-based2024.03 | 38.93 | — | — | — | |
| DialogueEINArchitecture Type=Sequence-based2024.03 | 38.92 | — | — | — | |
| SKAIGExtra Knowledge=COMET, Model type=GNN, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 38.88 | — | — | — | |
| DialogueCRNParameters=3.3M, Extra Knowledge=✗, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 38.79 | — | — | — | |
| SCCLArchitecture Type=Sequence-based2024.03 | 38.75 | — | — | — | |
| TODKATParameters=330M, Extra Knowledge=COMET, Model type=transformer, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 38.69 | — | — | — | |
| COMPMBase Model=COMPM, Grayscale Label Type=None2022.06 | 38.56 | — | — | — | |
| COSMICParameters=11.9M, Extra Knowledge=COMET, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 38.49 | — | — | — | |
| Psychological + SABase Model=Psychological, Grayscale Label Type=self-adjust2022.06 | 38.33 | — | — | — | |
| DAG-ERCParameters=9.5M, Extra Knowledge=✗, Model type=GNN, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 38.29 | — | — | — | |
| Llama2Parameters=12.5M(7B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + Backbone2023.09 | 38.21 | — | — | — | |
| COSMIC2021.06 | 38.11 | — | — | — | |
| COSMICArchitecture Type=Sequence-based2024.03 | 38.11 | — | — | — | |
| COSMICBackbone=-, External FREE=-, Training Protocol=Dataset-Specific Setting2026.04 | 38.11 | — | — | — | |
| DialogueGCNArchitecture Type=Graph-based2024.03 | 38.1 | — | — | — | |
| RoBERTa + FSABase Model=RoBERTa, Grayscale Label Type=future-self-adjust2022.06 | 38 | — | — | — | |
| LlamaParameters=12.5M(7B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + Backbone2023.09 | 37.98 | — | — | — | |
| RGATArchitecture Type=Graph-based2024.03 | 37.89 | — | — | — | |
| CauAINParameters=6.1M, Extra Knowledge=ATOMIC, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 37.87 | — | — | — | |
| RoBERTa + SABase Model=RoBERTa, Grayscale Label Type=self-adjust2022.06 | 37.69 | — | — | — | |
| ChatGLM2Parameters=12.5M(6B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + Backbone2023.09 | 37.69 | — | — | — | |
| DialogueRNNParameters=9.9M, Extra Knowledge=✗, Model type=recurrent, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 37.54 | — | — | — | |
| COG-BART + SABase Model=COG-BART, Grayscale Label Type=self-adjust2022.06 | 37.42 | — | — | — | |
| KATSBERTAverage of ten runs=true, knowledge base=SBERT2021.06 | 37.34 | 35.67 | — | — | |
| CoG-BARTParameters=415.1M, Extra Knowledge=✗, Model type=transformer, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 37.33 | — | — | — | |
| PsychologicalBase Model=Psychological, Grayscale Label Type=None2022.06 | 37.32 | — | — | — | |
| RoBERTa + CBase Model=RoBERTa, Grayscale Label Type=category2022.06 | 36.97 | — | — | — | |
| COG-BART + SBase Model=COG-BART, Grayscale Label Type=self2022.06 | 36.97 | — | — | — | |
| HiTrans2021.12 | 36.75 | — | — | — | |
| COG-BARTBase Model=COG-BART, Grayscale Label Type=None2022.06 | 36.48 | — | — | — | |
| DialogueGCNParameters=2.1M, Extra Knowledge=✗, Model type=GNN, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 36.43 | — | — | — | |
| RoBERTa (ours)Base Model=RoBERTa, Grayscale Label Type=None2022.06 | 36.3 | — | — | — | |
| RoBERTa + WBase Model=RoBERTa, Grayscale Label Type=word-embedding2022.06 | 36.05 | — | — | — | |
| BART-largeBackbone=BART-large2021.12 | 35.98 | 38.93 | — | — | |
| ChatGPTArchitecture Type=Sequence-based, Evaluation Protocol=3-shot2024.03 | 35.92 | — | — | — | |
| MTLParameters=1.2M, Extra Knowledge=✗, Model type=transformer, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 35.92 | — | — | — | |
| RoBERTaBackbone=RoBERTa2021.12 | 35.9 | 40.81 | — | — | |
| BERTBackbone=BERT2021.12 | 34.87 | 41.11 | — | — | |
| DialogXL2021.12 | 34.73 | — | — | — | |
| DialogueXLParameters=510M, Extra Knowledge=✗, Model type=transformer, Method Setting=Small-scale Pretrained Language Model2023.09 | 34.73 | — | — | — | |
| HiGRUEvaluation=Reproduced code2021.06 | 34.48 | 33.54 | — | — | |
| RGATParameters=13M, Extra Knowledge=✗, Model type=GNN, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 34.42 | — | — | — | |
| KET2021.06 | 34.39 | — | — | — | |
| KETParameters=2.6M, Extra Knowledge=ConceptNet, Model type=transformer, Method Setting=Small-scale Discriminant ERC-specific Model2023.09 | 34.39 | — | — | — | |
| DialogueGCNEvaluation=Reproduced code2021.06 | 34.29 | 33.13 | — | — | |
| XLNetBackbone=XLNet2021.12 | 34.13 | — | — | — | |
| TODKAT (-KB)Average of ten runs=true, knowledge base=false2021.06 | 33.79 | 32.62 | — | — | |
| DialogueRNNBackbone=GRU, External FREE=-, Training Protocol=Dataset-Specific Setting2026.04 | 31.7 | — | — | — | |
| ChatGLMParameters=12.5M(6B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + InstructERC2023.09 | 30.86 | — | — | — | |
| RoBERTa + SBase Model=RoBERTa, Grayscale Label Type=self2022.06 | 29.09 | — | — | — | |
| ChatGLMParameters=12.5M(6B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=LoRA + Backbone2023.09 | 25.71 | — | — | — | |
| ChatGLM2Parameters=12.5M(6B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=Zero-shot + InstructERC2023.09 | 24.4 | — | — | — | |
| PRC-EmoBackbone=Qwen3-8B, External FREE=NO, Training Protocol=Dataset-Unified Setting2026.04 | 21.94 | — | — | 22.53 | |
| Seed-1.8Backbone=Seed-1.8, External FREE=YES, Training Protocol=Dataset-Unified Setting2026.04 | 20.99 | — | — | 21.85 | |
| BiosERCBackbone=LLaMA2-13B, External FREE=NO, Training Protocol=Dataset-Unified Setting2026.04 | 20.55 | — | — | 19.83 | |
| ChatGLMParameters=12.5M(6B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=Zero-shot + InstructERC2023.09 | 19.6 | — | — | — | |
| DeepSeek-V3.2Backbone=DeepSeek-V3.2, External FREE=YES, Training Protocol=Dataset-Unified Setting2026.04 | 18.55 | — | — | 19.82 | |
| Llama2Parameters=12.5M(7B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=Zero-shot + InstructERC2023.09 | 8.36 | — | — | — | |
| LlamaParameters=12.5M(7B), Extra Knowledge=✗, Model type=LLM-based, Method Setting=Zero-shot + InstructERC2023.09 | 5.31 | — | — | — | |
| BERTTopic=false2023.07 | — | 41.11 | — | — | |
| COGBARTTopic=false2023.07 | — | 37.57 | — | — | |
| COSMIC2021.05 | — | — | 37.1 | — | |
| DAG-ERC2021.05 | — | — | 39.02 | — | |
| DAGNN2021.05 | — | — | 37.89 | — | |
| DialGCNTopic=false2023.07 | — | 33.13 | — | — | |
| DialogueGCNBackbone=RoBERTa2021.05 | — | — | 38.1 | — |