Nested Named Entity Recognition on ACE 2004 (test)
89.7F1 ScoreBINDER
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| BINDEREncoder=BERT-large/BioBERT-large2022.08 | 89.7 | 89.7 | 89.7 | — | |
| BINDEREncoder=BERT-base/BioBERT2022.08 | 88.7 | 88.3 | 89.1 | — | |
| TriaffineEncoder=ALBERT2021.10 | 88.56 | 88.88 | 88.24 | — | |
| DIFFUSIONNERCategory=Diffusion-based, Backbone=bert-large, K=60, T=10002023.05 | 88.39 | 88.11 | 88.66 | 85.62 | |
| DSpERTreporting_metric=best2022.10 | 88.31 | 88.29 | 88.32 | — | |
| PIQNBackbone=bert-large2022.03 | 88.14 | 88.48 | 87.81 | — | |
| Shen et al. (2022)2022.10 | 88.14 | 88.48 | 87.81 | — | |
| DSpERTreporting_metric=average2022.10 | 88.05 | 87.9 | 88.21 | — | |
| CNN_Nested_NERBackbone=ROBERTa-base, Parameters=125.6M, Data Source=Zhu and Li (2022)2022.08 | 88.03 | 87.77 | 88.28 | — | |
| BSBackbone=ROBERTa-base, Parameters=125.6M, Data Source=Zhu and Li (2022)2022.08 | 87.98 | 88.43 | 87.53 | — | |
| Zhu and Li (2022)2022.10 | 87.98 | 88.43 | 87.53 | — | |
| Zhu and Li (2022)Category=Span-based, Backbone=bert-large2023.05 | 87.98 | 88.43 | 87.53 | — | |
| Zhu and Li (2022)Backbone=BERT2024.06 | 87.98 | — | — | — | |
| Ours (8 bits)Backbone=BERT, Bits=82024.06 | 87.93 | — | — | — | |
| Nested NER as Latent Lexicalized Constituency Parsing2022.03 | 87.9 | 87.39 | 88.4 | — | |
| Lou et al. (2022)Category=Span-based, Backbone=bert-large2023.05 | 87.9 | 87.39 | 88.4 | — | |
| Lou et al. (2022)Backbone=BERT2024.06 | 87.9 | — | — | — | |
| Ours (10 bits)Backbone=BERT, Bits=102024.06 | 87.87 | — | — | — | |
| Pyramid-Fullextra_features=+ALBERT2022.03 | 87.74 | 87.71 | 87.78 | — | |
| PyramidEncoder=ALBERT2021.10 | 87.74 | 87.71 | 87.78 | — | |
| CNN_Nested_NERBackbone=BERT-large, Parameters=345.1M, Data Source=This Work (New Preprocessing)2022.08 | 87.74 | 87.98 | 87.5 | — | |
| CNN_Nested_NERBackbone=BERT-large, Parameters=345.1M, Data Source=Li et al. (2022)2022.08 | 87.61 | 87.82 | 87.4 | — | |
| PO-TreeCRFsrerun_on_current_data=true2022.03 | 87.6 | 87.62 | 87.57 | — | |
| W2NERBackbone=BERT-large, Parameters=355.4M, Data Source=Li et al. (2022)2022.08 | 87.52 | 87.33 | 87.71 | — | |
| Li et al. (2022)reporting_metric=average2022.10 | 87.52 | 87.33 | 87.71 | — | |
| Li et al. (2022a)Category=Span-based, Backbone=bert-large2023.05 | 87.52 | 87.33 | 87.71 | 85.23 | |
| Ours (12 bits)Backbone=BERT, Bits=122024.06 | 87.52 | — | — | — | |
| W2NERBackbone=BERT-large, Parameters=355.4M, Data Source=This Work (New Preprocessing), reproduction=true2022.08 | 87.43 | 87.17 | 87.7 | — | |
| Locate and LabelBackbone=BERT + GloVe2021.05 | 87.41 | 87.44 | 87.38 | — | |
| Shen et al.Encoder=BERT2021.10 | 87.41 | 87.44 | 87.38 | — | |
| Locate&Labelcontext_sentences=true, pos_tags=true, data_preprocessing=different2022.03 | 87.41 | 87.44 | 87.38 | — | |
| Shen et al. (2021a)2022.03 | 87.41 | 87.44 | 87.38 | — | |
| Locate and LabelEncoder=BERT2021.10 | 87.41 | 87.44 | 87.38 | — | |
| Shen et al. (2021)2022.10 | 87.41 | 87.44 | 87.38 | — | |
| Shen et al. (2021)Category=Span-based, Backbone=bert-large2023.05 | 87.41 | 87.44 | 87.38 | 84.87 | |
| Shen et al. (2021)Backbone=BERT2024.06 | 87.41 | — | — | — | |
| TriaffineEncoder=BERT2021.10 | 87.4 | 87.13 | 87.68 | — | |
| Yuan et al. (2022b)Category=Span-based, Backbone=bert-large2023.05 | 87.4 | 87.13 | 87.68 | 85.14 | |
| Ours (6 bits)Backbone=BERT, Bits=62024.06 | 87.39 | — | — | — | |
| CNN_Nested_NERBackbone=ROBERTa-base, Parameters=125.6M, Data Source=This Work (New Preprocessing)2022.08 | 87.31 | 87.33 | 87.29 | — | |
| Tan et al.Encoder=BERT-large/BioBERT2022.08 | 87.3 | 88.5 | 86.1 | — | |
| Tan et al. (2021)Backbone=BERT2024.06 | 87.28 | — | — | — | |
| Tan et al.Encoder=BERT2021.10 | 87.26 | 88.46 | 86.1 | — | |
| Seq2setcontext_sentences=true, pos_tags=true, data_preprocessing=different2022.03 | 87.26 | 88.46 | 86.1 | — | |
| Tan et al. (2021)2022.03 | 87.26 | 88.46 | 86.1 | — | |
| Sequence to SetEncoder=BERT2021.10 | 87.26 | 88.46 | 86.1 | — | |
| Tan et al. (2021)Category=Generation-based, Backbone=bert-large2023.05 | 87.26 | 88.46 | 86.1 | 84.91 | |
| BSBackbone=ROBERTa-base, Parameters=125.6M, Data Source=This Work (New Preprocessing), reproduction=true2022.08 | 87.08 | 87.32 | 86.84 | — | |
| CNN_Nested_NER (w.o. CNN)Backbone=ROBERTa-base, Parameters=125.2M, Data Source=Zhu and Li (2022)2022.08 | 87.05 | 86.71 | 87.4 | — | |
| Bottom-up Pointer NetworksEncoder=BERT2021.10 | 86.94 | 86.6 | 87.28 | — | |
| Locate&Labelrerun_on_current_data=true2022.03 | 86.94 | 87.27 | 86.61 | — | |
| Yang and Tu (2022)2022.03 | 86.94 | 86.6 | 87.28 | — | |
| Yang and Tu (2022)Backbone=BERT2024.06 | 86.94 | — | — | — | |
| Yan et al.Encoder=BERT2021.10 | 86.84 | 87.27 | 86.41 | — | |
| BARTNER2022.03 | 86.84 | 87.27 | 86.41 | — | |
| Yan et al. (2021)2022.03 | 86.84 | 87.27 | 86.41 | — | |
| BartNEREncoder=BART2021.10 | 86.84 | 87.27 | 86.41 | — | |
| Yan et al. (2021)2022.10 | 86.84 | 87.27 | 86.41 | — | |
| Yan et al. (2021b)Category=Generation-based, Backbone=bert-large2023.05 | 86.84 | 87.27 | 86.41 | 83.6 | |
| CNN_Nested_NER (w.o. CNN)Backbone=BERT-large, Parameters=343.6M, Data Source=Li et al. (2022)2022.08 | 86.81 | 86.54 | 87.09 | — | |
| Yan et al.Encoder=BART-large2022.08 | 86.8 | 87.3 | 86.4 | — | |
| Biaffine NER2020.05 | 86.7 | 87.3 | 86 | — | |
| Yu et al.2021.05 | 86.7 | 87.3 | 86 | — | |
| Yu et al. (2020)2022.03 | 86.7 | 87.3 | 86 | — | |
| BiaffineEncoder=BERT2021.10 | 86.7 | 87.3 | 86 | — | |
| Yu et al.Encoder=BERT-large/BioBERT2022.08 | 86.7 | 87.3 | 86 | — | |
| Yu et al. (2020)2022.10 | 86.7 | 87.3 | 86 | — | |
| Yu et al. (2020)Category=Span-based, Backbone=bert-large2023.05 | 86.7 | 87.3 | 86 | 84.2 | |
| Yu et al. (2020)Backbone=BERT2024.06 | 86.7 | — | — | — | |
| Seq2setrerun_on_current_data=true2022.03 | 86.65 | 87.05 | 86.26 | — | |
| CNN_Nested_NERBackbone=BERT-base, Parameters=110.5M, Data Source=Wan et al. (2022)2022.08 | 86.65 | 86.85 | 86.45 | — | |
| Fu et al.Encoder=BERT2021.10 | 86.6 | 86.7 | 86.5 | — | |
| PO-TreeCRFsdata_preprocessing=different2022.03 | 86.6 | 86.7 | 86.5 | — | |
| TreeCRFEncoder=BERT2021.10 | 86.6 | 86.7 | 86.5 | — | |
| Fu et al. (2021)Backbone=BERT2024.06 | 86.6 | — | — | — | |
| CNN_Nested_NER (w.o. CNN)Backbone=BERT-large, Parameters=343.6M, Data Source=This Work (New Preprocessing)2022.08 | 86.54 | 86.6 | 86.48 | — | |
| CNN_Nested_NER (w.o. CNN)Backbone=ROBERTa-base, Parameters=125.2M, Data Source=This Work (New Preprocessing)2022.08 | 86.48 | 86.09 | 86.88 | — | |
| BoningKnifePre-trained model=BERT-base2021.07 | 86.41 | 85.98 | 86.86 | — | |
| SGBackbone=BERT-base, Parameters=112.3M, Data Source=Wan et al. (2022)2022.08 | 86.31 | 86.7 | 85.93 | — | |
| Wan et al. (2022)Category=Span-based, Backbone=bert-large2023.05 | 86.31 | 86.7 | 85.93 | 83.57 | |
| Xu et al.Encoder=BERT2021.10 | 86.3 | 86.9 | 85.8 | — | |
| Wang et al.Encoder=BERT-large/BioBERT2022.08 | 86.3 | 86.1 | 86.5 | — | |
| Wan et al.Encoder=BERT-base2022.08 | 86.3 | 86.7 | 85.9 | — | |
| Xu et al. (2021)Backbone=BERT2024.06 | 86.3 | — | — | — | |
| Wang et al.suffix=2020a2021.05 | 86.28 | 86.08 | 86.48 | — | |
| Pyramid-Basic2022.03 | 86.28 | 86.08 | 86.48 | — | |
| Wang et al. (2020a)2022.03 | 86.28 | 86.08 | 86.48 | — | |
| PyramidEncoder=BERT2021.10 | 86.28 | 86.08 | 86.48 | — | |
| Wang et al. (2020)Category=Tagging-based, Backbone=bert-large2023.05 | 86.28 | 86.08 | 86.48 | 83.57 | |
| Wang et al. (2020)Backbone=BERT2024.06 | 86.28 | — | — | — | |
| Bottom-up Pointer Networks (w.o. E_yt-1)Encoder=BERT, Eq.1=w.o. E_yt-12021.10 | 86.27 | 85.66 | 86.88 | — | |
| Wang et al. (2020)Encoder=BERT2021.10 | 86.26 | 86.08 | 86.48 | — | |
| Wang et al. (2021)Backbone=BERT2024.06 | 86.06 | — | — | — | |
| Li et al. (2020b)2022.03 | 85.98 | 85.05 | 86.32 | — | |
| Li et al. (2020b)2022.10 | 85.98 | 85.05 | 86.32 | — | |
| Li et al. (2020)Category=Span-based, Backbone=bert-large2023.05 | 85.98 | 85.05 | 86.32 | 83.92 | |
| Nested Named Entity Recognition via Second-best Sequence Learning and DecodingBackbone=BERT+FLAIR2019.09 | 85.82 | 85.94 | 85.69 | — | |
| Second-bestEncoder=BERT2021.10 | 85.82 | 85.94 | 85.69 | — | |
| Ours (4 bits)Backbone=BERT, Bits=42024.06 | 85.81 | — | — | — | |
| Li et al.Encoder=BERT-large/BioBERT2022.08 | 85.8 | 85.8 | 85.8 | — |