Table 3 Results of relation mapping evaluation at the sentence level with CNN. (The average F1-scores after five times of random sub-sampling validation)
From: Deep learning of mutation-gene-drug relations from the literature
Relation | Word2vec | Without BEST Scores | With BEST Scores |
|---|---|---|---|
Mutation-Gene | None | 0.943 | 0.947 |
Google News | 0.954 | 0.955 | |
PubMed (Token-based) | 0.946 | 0.954 | |
PubMed (With BEST-EE) | 0.941 | 0.951 | |
Mutation-Drug | No word2vec | 0.803 | 0.820 |
Google News | 0.845 | 0.864 | |
PubMed (Token-based) | 0.829 | 0.841 | |
PubMed (With BEST-EE) | 0.837 | 0.856 |