Semisupervised Community Preserving Network Embedding with Pairwise Constraints

Table 5

AUC-ROC of link prediction (80% training set and 20% test set).

Methods	Polbooks	Cornell	Texas	Washington	Wisconsin	Email	Citeseer	Cora
DeepWalk	0.6257	0.8034	0.8539	0.6571	0.5340	0.6314	0.6436	0.5648
Line	0.7749	0.8356	0.8494	0.8276	0.7789	0.6685	0.6110	0.5920
node2vec	0.6747	0.8615	0.8408	0.7689	0.7650	0.6289	0.5442	0.5642
GraRep	0.9712	0.7864	0.8398	0.7643	0.8070	0.6614	0.6099	0.6563
MNMF	0.9922	0.9968	0.9707	0.9937	0.9911	0.9807	0.9907	0.9874
NECS	0.9993	0.8339	0.8675	0.8313	0.7949	0.7145	0.5718	0.6180
DNR (1)	0.9070	0.8758	0.8998	0.8634	0.8460	0.9134	0.7058	0.8103
DNR (2)	0.9013	0.8726	0.8937	0.8540	0.8484	0.8843	0.6378	0.7365
DNR (5)	0.8851	0.8697	0.8930	0.8479	0.8261	0.9011	0.5639	0.6718
DNR (10)	0.8831	0.8865	0.8582	0.8297	0.7976	0.9196	0.5511	0.6101
SMNMF (1)	0.9926	0.9934	0.9945	0.9978	0.9886	0.9725	0.9591	0.9629
SMNMF (2)	0.9846	0.9504	0.9848	0.9780	0.9829	0.9586	0.9419	0.9046
SMNMF (5)	0.9903	0.9518	0.9772	0.9774	0.9603	0.9273	0.9150	0.9408
SMNMF (10)	0.9753	0.9331	0.9448	0.9431	0.9255	0.8923	0.8825	0.9207