Table 1 The overall statistics of the CDR and CHR datasets
From: Biomedical relation extraction via knowledge-enhanced reading comprehension
Dataset | Splits | Documents | Chemical IDs | Disease IDs | pos | neg |
|---|---|---|---|---|---|---|
CDR | Training | 500 | 1479 | 1961 | 1038 | 4479 |
Development | 500 | 1519 | 1851 | 1012 | 4310 | |
Test | 500 | 1455 | 2007 | 1066 | 4471 | |
CHR | Training | 7298 | 28158 | – | 19643 | 69843 |
Development | 1182 | 4575 | – | 3185 | 11466 | |
Test | 3614 | 13800 | – | 9578 | 33339 |