Skip to main content

Advertisement

Table 4 Impact of separate BiLSTM layers in terms of the F1-score (%)

From: DTranNER: biomedical named entity recognition with deep learning-based label-label transition model

SettingsBC2GMBC5CDR-ChemicalBC5CDR-DiseaseNCBI-Disease
DTranNER-shared83.6993.5786.7588.01
DTranNER84.5694.1687.2288.62
  1. Note: “DTranNER-shared” is a variant model that shares the BiLSTM layer in “Unary-Network” and “Pairwise-Network.”