Table 4.
Statistics of the datasets used in the experiments
| CDR Disease | CDR Chem | CT Condition | CT Intervention | |
|---|---|---|---|---|
| Domain | Abstracts | Abstracts | Clinical trials | Clinical trials |
| Entity type | Disease | Chemicals | Conditions | Drugs |
| Terminology | MEDIC | CTD Chemicals | MeSH | Drugbank |
| Entity level statistics | ||||
| % numerals | 0.11% | 7.32% | 7.69% | 25.3% |
| % punctuation | 1.21% | 0.07% | 14.28% | 24.83% |
| Avg. len | 14.88 | 11.27 | 17.92 | 21.68 |
| Number of pre-processed entity mentions | ||||
| Train set | 4182 | 5203 | – | – |
| Dev set | 4244 | 5347 | 100 | 100 |
| Test set | 4424 | 5385 | 719 | 975 |
| Filtered test | 1240 (28.02%) | 826 (15.38%) | 642 (78.4%) | 846 (78.7%) |
Note: Two sets of annotated clinical trials’ fields are marked with ‘CT’.