Skip to main content


Table 2 Comparison with BioCreative II best result.

From: Extract interaction detection methods from the biological literature

  Precision Recall F-Score
BioCreative II Best Run 0.506 0.522 0.483
CMW model 0.654 0.545 0.543
improvement +29.2% +4.4% +12.4%
  1. We operate the CMW model on the BioCreative II testing corpus (300 full text documents) to compare with the best result reported in the evaluation. We set the topic size to be 300 according to the result in Figure 5. The CMW model achieved competitive results, F-Score improved 12.4%.