Table 4 Precision (Pr), Recall (Re), and F1 measure of the models (BioBERT, CNN, decision tree) and baselines (sentence- and abstract-level) calculated on the concept-level test set
From: An annotated dataset for extracting gene-melanoma relations from scientific literature
Models | Pr | Re | F1 |
|---|---|---|---|
BioBERT | 74.42 | 77.29(67.09) | 75.83 |
CNN | 70.50 | 71.01(59.15) | 70.76 |
decision tree | 66.27 | 67.87(44.18) | 67.06 |
sentence-level | 49.68 | 94.93 | 65.23 |
abstract-level | 35.75 | 100.00 | 52.67 |