Span SRL results on the CoNLL 2005 English WSJ test set, in terms of constituent syntax F1 score (Syn-F1), semantic precision (P), semantic recall (R), semantic labeled F1 score (Sem-F1), and the ratio Sem-F1/Syn-F1.
| System . | . | PLM . | Syn-F1 . | P . | R . | Sem-F1 . | Sem-F1/Syn-F1 . |
|---|---|---|---|---|---|---|---|
| (He et al. 2017) | Choe&Charniak Parser | 93.8 | − | − | 84.8 | 90.41 | |
| Gold syntax | 100.0 | − | − | 87.0 | 87.00 | ||
| (Wang et al. 2019) | Kitaev&Klein Parser | +E | 95.4 | − | − | 88.2 | 92.45 |
| Gold syntax | +E | 100.0 | − | − | 92.2 | 92.20 | |
| (Marcheggiani and Titov 2020) | Kitaev&Klein Parser | 95.4 | 85.8 | 85.1 | 85.4 | 89.52 | |
| Sequence-based + Syntax GCN encoder | Choe&Charniak Parser | 93.8 | 86.4 | 84.8 | 85.6 | 91.26 | |
| Choe&Charniak Parser | +E | 93.8 | 87.2 | 86.8 | 87.0 | 92.75 | |
| Kitaev&Klein Parser | +E | 95.4 | 88.3 | 89.1 | 88.5 | 92.77 | |
| Gold syntax | +E | 100.0 | 93.2 | 92.4 | 92.6 | 92.60 | |
| System . | . | PLM . | Syn-F1 . | P . | R . | Sem-F1 . | Sem-F1/Syn-F1 . |
|---|---|---|---|---|---|---|---|
| (He et al. 2017) | Choe&Charniak Parser | 93.8 | − | − | 84.8 | 90.41 | |
| Gold syntax | 100.0 | − | − | 87.0 | 87.00 | ||
| (Wang et al. 2019) | Kitaev&Klein Parser | +E | 95.4 | − | − | 88.2 | 92.45 |
| Gold syntax | +E | 100.0 | − | − | 92.2 | 92.20 | |
| (Marcheggiani and Titov 2020) | Kitaev&Klein Parser | 95.4 | 85.8 | 85.1 | 85.4 | 89.52 | |
| Sequence-based + Syntax GCN encoder | Choe&Charniak Parser | 93.8 | 86.4 | 84.8 | 85.6 | 91.26 | |
| Choe&Charniak Parser | +E | 93.8 | 87.2 | 86.8 | 87.0 | 92.75 | |
| Kitaev&Klein Parser | +E | 95.4 | 88.3 | 89.1 | 88.5 | 92.77 | |
| Gold syntax | +E | 100.0 | 93.2 | 92.4 | 92.6 | 92.60 | |