Biomedical relation extraction via knowledge-enhanced reading comprehension

BMC Bioinformatics

Table 3 Doc-level performance comparison over our proposed model without knowledge on the CHR dataset

	Model	P (%)	R (%)	F1 (%)
NN-based ML
Relation classification	CNN-RE [2]	81.2	87.3	84.1
	RNN-RE [2]	83.0	90.1	86.4
	GCNN [2]	84.7	90.5	87.5
Reading comprehension	RC (Ours)	93.5	93.0	93.3

ISSN: 1471-2105