Table 9 The recognizing performance on the test dataset of the CDR corpus when different training sets were applied without postprocessing
From: A document level neural model integrated domain knowledge for chemical-induced disease relations
Training set | Knowledge | P(%) | R(%) | F(%) |
|---|---|---|---|---|
Only original training set | Without KB | 45.2 | 68.1 | 54.3 |
With KB | 55.8 | 81.4 | 66.1 | |
Train +development set | Without KB | 54.3 | 65.9 | 59.5 |
With KB | 61.5 | 78.7 | 69.0 |