Skip to main content

Table 8 Evaluation results on the Bacteria_Corpus, using a 0.5 threshold. We evaluated three models: majority voting, generative model, and discriminative model

From: Extracting chemical reactions from text using Snorkel

ModelCoveragePrecisionRecallF1 Score
Majority voting0.921.000.200.34
Generative model0.941.000.330.49
Discriminative model1.000.500.610.55
\