Argument identification and classification on CoNLL 2009 (test)
90.7F1 ScoreHigh-Order model
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| High-Order modelmode=Pipeline, predicates=Gold2021.01 | 90.7 | 90.3 | 91 | |
| Li et al. (2019)mode=Pipeline, predicates=Gold2021.01 | 90.5 | 90.9 | 90.2 | |
| BERT-LSTM-largeBackbone=BERT-large2019.04 | 90.3 | — | — | |
| Cai et al. (2018)mode=Pipeline, predicates=Gold2021.01 | 90 | 89.9 | 90.2 | |
| BERT-LSTM-baseBackbone=BERT-base2019.04 | 89.8 | — | — | |
| He et al. (2018)mode=Pipeline, predicates=Gold2021.01 | 89.7 | 89.8 | 89.6 | |
| Marcheggiani and Titov (2017)mode=Pipeline, predicates=Gold2021.01 | 88 | 89.1 | 86.8 | |
| Shi and Zhang2019.04 | 87.1 | — | — | |
| FitzGerald et al. (2015)mode=Pipeline, predicates=Gold2021.01 | 86.7 | — | — | |
| Roth and Lapata (2016)mode=Pipeline, predicates=Gold2021.01 | 86.7 | 88.1 | 85.3 | |
| Björkelund et al. (2010)mode=Pipeline, predicates=Gold2021.01 | 85.8 | 87.1 | 84.5 | |
| Zhao et al. (2009)mode=Pipeline, predicates=Gold2021.01 | 85.4 | — | — |