Semisupervised Community Preserving Network Embedding with Pairwise Constraints

Table 4

AUC-ROC of link prediction (90% training set and 10% test set).

Methods	Polbooks	Cornell	Texas	Washington	Wisconsin	Email	Citeseer	Cora
DeepWalk	0.6266	0.8917	0.8757	0.6776	0.5381	0.6363	0.6611	0.5849
Line	0.8769	0.9203	0.8979	0.8621	0.8649	0.6664	0.6535	0.6282
node2vec	0.5745	0.9046	0.9481	0.8565	0.8112	0.6319	0.5514	0.5892
GraRep	0.9615	0.8481	0.8979	0.8351	0.8638	0.6613	0.6487	0.6754
MNMF	0.9941	0.9967	0.9978	0.9983	0.9907	0.9822	0.9916	0.9806
NECS	0.9998	0.8988	0.8942	0.8710	0.8445	0.8132	0.5972	0.6347
DNR (1)	0.9016	0.9070	0.9224	0.8566	0.8724	0.9176	0.7412	0.8367
DNR (2)	0.8946	0.9034	0.9133	0.8410	0.8730	0.8905	0.6191	0.7520
DNR (5)	0.8944	0.8979	0.9142	0.8171	0.8662	0.9053	0.5192	0.6881
DNR (10)	0.8907	0.9160	0.8466	0.7797	0.8314	0.9244	0.5510	0.6362
SMNMF (1)	0.9884	0.9956	0.9999	0.9882	0.9851	0.9745	0.9630	0.9586
SMNMF (2)	0.9963	0.9890	0.9972	0.9940	0.9877	0.9616	0.9416	0.8889
SMNMF (5)	0.9912	0.9787	0.9546	0.9660	0.9787	0.9293	0.9187	0.9323
SMNMF (10)	0.9914	0.9556	0.9653	0.9299	0.9525	0.9009	0.8972	0.9177