To evaluate the performance of the extraction model, we utilized the TAC evaluation scripts.
Table 1. Performance metrics evaluated against the TAC gold standard
| Metric | TAC (Best Model†) | SIDER 4.1 | OnSIDES v1.0.0 | OnSIDES v2/3.0.0 |
|---|---|---|---|---|
| F1 Score | 82.19 | 74.36 | 82.01 | 87.54 |
| Precision | 80.69 | 43.49 | 88.76 | 91.29 |
| Recall | 85.05 | 52.89 | 77.12 | 84.08 |
† Roberts, Demner-Fushman, & Tonning, Overview of the TAC 2017