Constituent Parsing on CTB (test)
93.59F1 Scoreattach-juxtapose (BERT)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| attach-juxtapose (BERT)Pre-training=BERT, Extra supervision=false2020.10 | 93.59 | 93.4 | 93.8 | 49.72 | |
| Mrini et al.Pre-training=BERT, Extra supervision=true2020.10 | 92.64 | 91.85 | 93.45 | — | |
| LAL-Parser2019.11 | 92.64 | 91.85 | 93.45 | — | |
| Joint Span HPSGDecoding method=Joint, Pre-trained=RoBERTa2019.07 | 92.55 | 92.5 | 92.61 | — | |
| Zhou and ZhaoPre-training=None, Extra supervision=true2020.10 | 92.18 | 92.03 | 92.33 | — | |
| Zhou and Zhao (2019)Backbone=BERT2019.11 | 92.18 | 92.03 | 92.33 | — | |
| Kitaev et al.Pre-training=BERT, Extra supervision=false, implementation_source=Fried et al. [13]2020.10 | 92.14 | — | — | 44.42 | |
| Liu and ZhangPre-training=BERT, Extra supervision=false, implementation_source=Fried et al. [13]2020.10 | 91.81 | — | — | 44.94 | |
| Kitaev and Klein (2018b)2019.07 | 91.75 | 91.55 | 91.96 | — | |
| Kitaev et al.Pre-training=None, Extra supervision=false2020.10 | 91.75 | 91.55 | 91.96 | — | |
| Kitaev et al. (2018)2019.11 | 91.75 | 91.55 | 91.96 | — | |
| Joint Span HPSGDecoding method=Joint, Pre-trained=RoBERTa2019.07 | 90.02 | 90.14 | 89.89 | — | |
| Joint Span HPSGDecoding method=Joint2019.07 | 89.4 | 89.09 | 89.7 | — | |
| Division Span HPSGDecoding method=Division2019.07 | 89.27 | 88.75 | 89.79 | — | |
| In-order parsermode=rerank2017.07 | 88 | — | — | — | |
| Joint Span HPSGDecoding method=Joint2019.07 | 87.92 | 87.87 | 87.96 | — | |
| Division Span HPSGDecoding method=Division2019.07 | 87.74 | 87.41 | 88.07 | — | |
| Bottom-up parsermode=rerank2017.07 | 87.5 | — | — | — | |
| Teng and Zhang (2018)2019.07 | 87.3 | 87.1 | 87.5 | — | |
| Teng and Zhang (2018)2019.11 | 87.3 | 86.6 | 88 | — | |
| Fried and Klein (2018)2019.07 | 87 | — | — | — | |
| Fried and Klein (2018)2019.11 | 87 | — | — | — | |
| Dyer et al. (2016)mode=rerank2017.07 | 86.9 | — | — | — | |
| Top-down parsermode=rerank2017.07 | 86.9 | — | — | — | |
| Wang et al. (2015)supervision=semi-supervision2017.07 | 86.6 | — | — | — | |
| Shen et al. (2018)2019.07 | 86.5 | 86.6 | 86.4 | — | |
| Shen et al. (2018)2019.11 | 86.5 | 86.6 | 86.4 | — | |
| Wang and Xue (2014)supervision=semi-supervision2017.07 | 86.3 | — | — | — | |
| In-order parsersupervision=fully-supervision2017.07 | 86.1 | — | — | — | |
| Liu and Zhang (2017a)2019.07 | 86.1 | — | — | — | |
| Liu and ZhangPre-training=None, Extra supervision=false2020.10 | 86.1 | — | — | — | |
| Liu and Zhang (2017a)2019.11 | 86.1 | — | — | — | |
| Bottom-up parsersupervision=fully-supervision2017.07 | 85.7 | — | — | — | |
| Zhu et al. (2013)supervision=semi-supervision2017.07 | 85.6 | — | — | — | |
| Liu and Zhang (2017)supervision=fully-supervision2017.07 | 85.5 | — | — | — | |
| Liu and Zhang (2017b)2019.07 | 85.5 | 85.9 | 85.2 | — | |
| Liu and Zhang (2017b)2019.11 | 85.5 | 85.9 | 85.2 | — | |
| Dyer et al. (2016)supervision=fully-supervision2017.07 | 84.6 | — | — | — | |
| Top-down parsersupervision=fully-supervision2017.07 | 84.6 | — | — | — | |
| Dyer et al. (2016)2019.07 | 84.6 | — | — | — | |
| Dyer et al. (2016)2019.11 | 84.6 | — | — | — | |
| Zhu et al. (2013)supervision=fully-supervision2017.07 | 83.2 | — | — | — | |
| Wang et al. (2015)supervision=fully-supervision2017.07 | 83.2 | — | — | — | |
| Wang et al. (2015)2019.07 | 83.2 | — | — | — | |
| Charniak and Johnson (2005)mode=rerank2017.07 | 82.3 | — | — | — |