Table 7. Classification results of DDI classification.

SOTA: state-of-the-art. P: Precision. R: Recall. The SOTA results are extracted from [63]. Direct comparison: the results of embeddings using the same method (cbow) and same hyperparameters. Different embedding methods: the results of BioConceptVec (skip-gram), BioConceptVec (GloVe) and BioConceptVec (fastText). The highest results of each section are marked as bold.

Model	F1-score on each relation type				Overall performance
	Int	Advice	Effect	Mechanism	P	R	F
Zhang et al. (SOTA)	0.5400	0.8000	0.7200	0.7400	0.7400	0.7200	0.7300
SEN	0.3569	0.8336	0.7978	0.8463	0.7940	0.7832	0.7776
Direct comparison
SEN + BioAvgWord (cbow)	0.3150	0.7787	0.8000	0.8824	0.7883	0.7814	0.7731
SEN     + Yu et al. (cbow)	0.4285	0.8263	0.8133	0.8559	0.7948	0.7961	0.7916
SEN     + BioConceptVec (cbow)	0.5206	0.8423	0.8191	0.8692	0.8167	0.8161	0.8105
Different embedding methods
SEN    + BioConcept (skip-gram)	0.4090	0.8164	0.8255	0.8626	0.8088	0.8025	0.7941
SEN    + BioConcept (GloVe)	0.4587	0.8100	0.8160	0.8702	0.8046	0.8029	0.7963
SEN    + BioConcept (fastText)	0.4382	0.8153	0.8200	0.8571	0.7999	0.7998	0.7930