From: Unsupervised inference of implicit biomedical events using context triggers
Existing models | Â | Â | F1 | Recall | Precision | |
BB-event task participants (2016) | LIMSI | 48.5 | 64.6 | 38.8 | ||
 | TurkuNLP | 52.1 | 44.8 | 62.3 | ||
 | VERSE | 55.8 | 61.5 | 51.0 | ||
State-of-the-art systems | Li et al. [23] | 58.1 | 58.0 | 56.3 | ||
 | Li et al. [24] | 57.4 | 56.8 | 59.4 | ||
 | Gupta et al. [20] | 58.7 | 65.7 | 53.0 | ||
Proposed models | #intra | #cross | F1 | Recall | Precision | |
M1: Intra-clause syntactic patterns | 213 | 0 | 37.7 | 30.7 | 48.8 | |
M2: Intra-clause syntactic patterns + trigger-based inference (train) | 246 | 0 | 40.4 | 34.8 | 48.1 | |
M3: Intra-clause syntactic patterns + trigger-based inference (unlabeled) | 417 | 64 | 56.7 | 68.6 | 48.2 | |
M4: VERSE (2016) + trigger-based inference (unlabeled) | 339 | 54 | 58.9 | 63.8 | 54.9 | |
M5: VERSE (2016) + trigger-based inference (unlabeled) (without linguistic modality detection) | 339 | 63 | 58.3 | 63.8 | 53.6 |