Skip to main content

Table 1 Precision-at-2 (P@2) of identifying target disease/trait mention of a GWAS study

From: Weakly supervised learning of biomedical information extraction from curated data

Method Arithmetic Harmonic
Cost-insensitive 68.65 % 79.62 %
Cost-sensitive 78.05 % 87.46 %
BIOADI+Cost-sensitive 75.57 % 87.29 %
CRF+Cost-sensitive 65.79 % 75.24 %