Table 1.
Statistics of the data set. Rare words are words that occur only once in the data. Unknown words refer to words that are not seen in the training set
| Item | Training | Development |
|---|---|---|
| Document | 242 | 61 |
| Entities | 41 171 | 9776 |
| Nest level 1 entity (flat entities) | 41 109 | 9760 |
| Nest level 2 entity | 61 | 16 |
| Nest level 3 entity | 1 | 0 |
| Polysemous entity | 47 | 13 |
| Textually nested entity | 15 | 3 |
| ADE | 785 | 174 |
| Dosage | 3401 | 820 |
| Drug | 13 109 | 3114 |
| Duration | 499 | 93 |
| Form | 5340 | 1311 |
| Frequency | 5075 | 1205 |
| Reason | 3105 | 750 |
| Route | 4479 | 996 |
| Strength | 5378 | 1313 |
| Unknown words /Unique words | – | 17.00% |
| Rare words /Unique words | 37.19% | 37.69% |
| EUNKs/All entities | – | 2.67% |
| ERAREs /All entities | 1.89% | 3.88% |
Abbreviations: ADE, adverse drug event; EUNKs, entities that contain unknown words, ERAREs, entities that contain rare words.