Named Entity Recognition on OntoNotes 4.0 (test)
89.52F1 ScoreContext-to-Vec
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Context-to-Vecbackbone=BERT2022.10 | 89.52 | — | — | 92.8 | |
| GloVe2022.10 | 89.13 | — | — | 91.84 | |
| BERT+Skip-gram2022.10 | 88.98 | — | — | 92.3 | |
| SynGCN2022.10 | 88.75 | — | — | 92.23 | |
| BERTavgpooling=average2022.10 | 84.51 | — | — | 90.76 | |
| Skip-gram2022.10 | 83.9 | — | — | 89.03 | |
| W2NER2021.12 | 83.08 | 82.31 | 83.36 | — | |
| SoftLexicon (LSTM) + BERTInput=No seg2019.08 | 82.81 | 83.41 | 82.21 | — | |
| Ma et al. (2020)2021.12 | 82.81 | 83.41 | 82.21 | — | |
| BABERT-LECategory=Supervised Lexicon Enhanced Model, External Knowledge=Yes2022.10 | 82.35 | — | — | — | |
| ChineseBERTModel scale=Large2021.06 | 82.18 | 80.77 | 83.65 | — | |
| BABERTCategory=Pre-Trained Language Model2022.10 | 81.9 | — | — | — | |
| BERT + LSTM + CRFInput=No seg2019.08 | 81.82 | 81.99 | 81.65 | — | |
| Li et al. (2020b)2021.12 | 81.82 | — | — | — | |
| NEZHACategory=Pre-Trained Language Model2022.10 | 81.74 | — | — | — | |
| ChineseBERTModel scale=Base2021.06 | 81.65 | 80.03 | 83.33 | — | |
| BERT-LECategory=Supervised Lexicon Enhanced Model, External Knowledge=Yes, Reproduced=Yes2022.10 | 81.59 | — | — | — | |
| RoBERTaModel scale=Large2021.06 | 81.39 | 80.72 | 82.07 | — | |
| BERTCategory=Pre-Trained Language Model2022.10 | 80.98 | — | — | — | |
| ERNIE-GramCategory=Pre-Trained Language Model2022.10 | 80.96 | — | — | — | |
| BERTModel scale=Base2021.06 | 80.87 | 79.69 | 82.09 | — | |
| BERT-wwmCategory=Pre-Trained Language Model2022.10 | 80.87 | — | — | — | |
| ERNIECategory=Pre-Trained Language Model2022.10 | 80.38 | — | — | — | |
| RoBERTaModel scale=Base2021.06 | 80.37 | 80.43 | 80.3 | — | |
| ZENCategory=Pre-Trained Language Model2022.10 | 80.06 | — | — | — | |
| BERT-TaggerInput=No seg2019.08 | 77.93 | 76.01 | 79.96 | — | |
| Yang et al. 2016variant=*2019.04 | 76.4 | 72.98 | 80.15 | — | |
| Yang et al., 2016*†Input=Gold seg2019.08 | 76.4 | 72.98 | 80.15 | — | |
| SoftLexicon (LSTM) + bicharInput=No seg2019.08 | 76.16 | 77.13 | 75.22 | — | |
| Word-based (LSTM) + char + bicharInput=Gold seg2019.08 | 75.77 | 78.62 | 73.13 | — | |
| SoftLexicon (LSTM)Input=No seg2019.08 | 75.64 | 77.28 | 74.07 | — | |
| Che et al. 2013variant=*2019.04 | 75.02 | 77.71 | 72.51 | — | |
| Che et al., 2013*Input=Gold seg2019.08 | 75.02 | 77.71 | 72.51 | — | |
| LR-CNN (Gui et al., 2019)Input=No seg2019.08 | 74.45 | 76.4 | 72.6 | — | |
| Gui et al. (2019)2021.12 | 74.45 | 76.4 | 72.6 | — | |
| Wang et al. 2013variant=*2019.04 | 74.32 | 76.43 | 72.32 | — | |
| Wang et al., 2013*Input=Gold seg2019.08 | 74.32 | 76.43 | 72.32 | — | |
| Zhang and Yang 2018variant=†2019.04 | 73.88 | 76.35 | 71.56 | — | |
| Lattice-LSTMInput=No seg2019.08 | 73.88 | 76.35 | 71.56 | — | |
| Zhang and Yang (2018)2021.12 | 73.88 | 76.35 | 71.56 | — | |
| LatticeCategory=Supervised Lexicon Enhanced Model, External Knowledge=Yes2022.10 | 73.88 | — | — | — | |
| CAN Model2019.04 | 73.64 | 75.05 | 72.29 | — | |
| Yan et al. (2019)2021.12 | 72.43 | — | — | — | |
| Char-based (LSTM) + bichar + ExSoftwordInput=No seg2019.08 | 72.4 | 73.8 | 71.05 | — | |
| Baseline + CNN2019.04 | 72.1 | 72.69 | 71.51 | — | |
| Char-based (LSTM) + bichar + softwordInput=No seg2019.08 | 71.89 | 74.36 | 69.43 | — | |
| Zhang and Yang 2018variant=+2019.04 | 71.81 | 74.36 | 69.43 | — | |
| Word-based (LSTM) + char + bicharInput=Auto seg2019.08 | 71.7 | 73.36 | 70.12 | — | |
| Baseline2019.04 | 71.15 | 70.67 | 71.64 | — | |
| Word-based (LSTM)Input=Gold seg2019.08 | 69.52 | 76.66 | 63.6 | — | |
| Yang et al. 20162019.04 | 68.57 | 65.59 | 71.84 | — | |
| Yang et al., 2016Input=Gold seg2019.08 | 68.57 | 65.59 | 71.84 | — | |
| Char-based (LSTM) + ExSoftwordInput=No seg2019.08 | 68.13 | 69.9 | 66.46 | — | |
| Word-based (LSTM)Input=Auto seg2019.08 | 65.63 | 72.84 | 59.72 | — | |
| Char-based (LSTM)Input=No seg2019.08 | 64.3 | 68.79 | 60.35 | — |