Table 3.

Comparison of ADR classification precision (P), recall (R), and F-measure (F) of ADRMine with embedding cluster features (ADRMine_{WITH_CLUSTER}) and the baselines systems on two different corpora: DS and Twitter

Method	DS			Twitter
	P	R	F	P	R	F
MetaMapADR_LEXICON	0.470	0.392	0.428	0.394	0.309	0.347
MetaMapSEMANTIC_TYPE	0.289	0.484	0.362	0.230	0.403	0.293
Lexicon-based	0.577	0.724	0.642	0.561	0.610	0.585
SVM	0.869	0.671	0.760	0.778	0.495	0.605
ADRMineWITHOUT_CLUSTER	0.874	0.723	0.791	0.788	0.549	0.647
ADRMineWITH_CLUSTER	0.860	0.784	0.821	0.765	0.682	0.721

The highest values in each column are highlighted in bold.