While syntactic dependency annotations concentrate on the surface or functional structure of a sentence, semantic dependency annotations aim to capture between-word relationships that are more closely related to the meaning of a sentence, using graph-structured representations. We extend the LSTM-based syntactic parser of Dozat and Manning (2017) to train on and generate these graph structures. The resulting system on its own achieves state-of-the-art performance, beating the previous, substantially more complex state-of-the-art system by 0.6% labeled F1. Adding linguistically richer input representations pushes the margin even higher, allowing us to beat it by 1.9% labeled F1.
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Semantic Parsing | DM | In-domain | 93.7 | Dozat et al. (2018) |
| Semantic Parsing | DM | Out-of-domain | 88.9 | Dozat et al. (2018) |
| Semantic Parsing | PSD | In-domain | 81 | Dozat et al. (2018) |
| Semantic Parsing | PSD | Out-of-domain | 79.4 | Dozat et al. (2018) |
| Semantic Parsing | PAS | In-domain | 93.9 | Dozat et al. (2018) |
| Semantic Parsing | PAS | Out-of-domain | 90.6 | Dozat et al. (2018) |