Comparing a knowledge-driven approach to a supervised machine learning approach in large-scale extraction of drug-side effect relationships from free-text biomedical literature

Table 1 Compare knowledge-driven approach (KD) to SVM for ten drugs

Drug	KD			SVM			F1 Increase
	Precision	Recall	F1	Precision	Recall	F1
Fluoxetine	0.370	0.425	0.395	0.178	0.838	0.294	34.4%
Tacrolimus	0.315	0.519	0.392	0.140	0.961	0.245	60.0%
Risperidone	0.561	0.622	0.590	0.234	0.959	0.377	56.5%
Carbamazepine	0.341	0.627	0.442	0.118	0.896	0.209	111.5%
Ibuprofen	0.230	0.406	0.294	0.106	0.891	0.189	55.6%
Olanzapine	0.463	0.492	0.477	0.180	0.921	0.301	58.5%
Morphine	0.214	0.611	0.317	0.090	0.833	0.163	94.5%
Phenytoin	0.229	0.403	0.292	0.104	0.881	0.186	57.0%
Methotrexate	0.230	0.607	0.333	0.082	0.984	0.152	119.1%
Ciprofloxacin	0.397	0.377	0.387	0.121	0.836	0.212	82.5%
Average	0.335	0.509	0.392	0.135	0.900	0.233	73.0%

ISSN: 1471-2105