Skip to main content

Table 5 Average precision rates when adding grammar relations to single words.

From: Classifying protein-protein interaction articles using word and syntactic features

Feature Set Naïve Bayes SVM Huber
Single Words (SW) 0.6169 0.6600 0.6646
Grammar Relations (GR) 0.6281 0.6391 0.6417
SW + GR 0.6538 0.6726 0.6771
  1. The best score is obtained by using both single words and grammar relations for all classifiers. The used training data was BioCreative II, BioCreative II.5, and BioCreative III training corpora. The performance was measured for the BioCreative development set.