Context Attention Heterogeneous Network Embedding

Table 2

AUC scores on HEP-TH.

% training edges	15%	25%	35%	45%	55%	65%	75%	85%	95%
DeepWalk	0.658	0.757	0.819	0.864	0.891	0.898	0.897	0.919	0.912
LINE	0.500	0.594	0.717	0.755	0.788	0.806	0.846	0.839	0.923
Node2vec	0.663	0.776	0.845	0.866	0.884	0.899	0.906	0.929	0.915
GraRep	0.628	0.735	0.776	0.841	0.853	0.872	0.885	0.896	0.914
Naive Combination	0.766	0.782	0.788	0.802	0.827	0.856	0.883	0.912	0.928
TADW	0.806	0.818	0.857	0.893	0.902	0.918	0.924	0.936	0.948
TENE	0.778	0.807	0.839	0.862	0.899	0.923	0.928	0.938	0.939
ASNE	0.783	0.802	0.833	0.869	0.893	0.905	0.918	0.926	0.938
CAHNE(w/o context)	0.730	0.796	0.854	0.894	0.893	0.913	0.916	0.921	0.923
CAHNE	0.786	0.818	0.860	0.896	0.902	0.928	0.935	0.937	0.954
CAHNE-a	0.858	0.854	0.869	0.898	0.910	0.929	0.941	0.945	0.977