edge2vec: Representation learning using edge semantics for biomedical knowledge discovery

BMC Bioinformatics

Table 4 Searching accuracy for retrieving potential compound-gene bindings

algorithm	P@10	P@100	Recall@10	Recall@100	MAP	NDCG	MRR
DeepWalk	0.0623	0.0198	0.0725	0.2780	0.0707	0.1444	0.1502
LINE	0.0186	0.0069	0.0109	0.0360	0.0042	0.0234	0.0532
node2vec	0.0714	0.0277	0.0859	0.2804	0.0786	0.1690	0.1676
metapath2vec (Co-Dr-Ge-Dr-Co)	0.0000	0.0001	0.0000	0.0011	0.0000	0.0004	0.0001
metapath2vec++ (Co-Dr-Ge-Dr-Co)	0.0157	0.0039	0.0082	0.0200	0.0040	0.0172	0.0509
edge2vec	0.0843*	0.0329*	0.0990*	0.3092*	0.0809*	0.1840*	0.1882*

Symbol “*” highlights the cases where our model significantly beats the best baseline with p value smaller than 0.01

ISSN: 1471-2105