Explicit Discourse Relation Recognition on PDTB 2.0 (test)
96.73AccuracyDiscoPrompt (T5-large)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DiscoPrompt (T5-large)Connective handling=w/ Gold Connective, Backbone=T5-large2023.05 | 96.73 | 95.64 | |
| Varia et al. (2019)2023.05 | 96.2 | 95.48 | |
| Fine-tuning (T5-large)Connective handling=w/ Gold Connective, Backbone=T5-large2023.05 | 95.41 | 94.94 | |
| Dai and Huang (2019)2023.05 | 95.39 | 94.84 | |
| Zhou et al. (2022)2023.05 | 94.78 | 93.59 | |
| Dai and Huang (2018)2023.05 | 94.46 | 93.7 | |
| Pitler and Nenkova (2009)2023.05 | 94.15 | — | |
| DiscoPrompt (T5-large)Connective handling=w/ Connective Mask, Backbone=T5-large2023.05 | 78.35 | 74.62 | |
| Fine-tuning (T5-large)Connective handling=w/o Connective, Backbone=T5-large2023.05 | 74.47 | 72.38 |