Semisupervised Community Preserving Network Embedding with Pairwise Constraints

Table 6

AUC-ROC of link prediction (70% training set and 30% test set).

Methods	Polbooks	Cornell	Texas	Washington	Wisconsin	Email	Citeseer	Cora
DeepWalk	0.6146	0.7829	0.7621	0.6374	0.5612	0.6322	0.6203	0.5568
Line	0.7750	0.8221	0.8112	0.7858	0.7651	0.6703	0.6171	0.5902
node2vec	0.6411	0.8359	0.8372	0.7398	0.6430	0.6309	0.5451	0.5395
GraRep	0.9640	0.6956	0.8159	0.7440	0.7762	0.6623	0.6179	0.6450
MNMF	0.9900	0.9987	0.9804	0.9960	0.9985	0.9820	0.9910	0.9855
NECS	0.9974	0.8021	0.8148	0.7839	0.7549	0.7135	0.5762	0.6150
DNR (1)	0.8914	0.8802	0.8718	0.8369	0.8720	0.9118	0.6911	0.7884
DNR (2)	0.8944	0.8760	0.8700	0.8482	0.8736	0.8843	0.6174	0.7182
DNR (5)	0.8744	0.8761	0.8750	0.8307	0.8616	0.9012	0.5455	0.6790
DNR (10)	0.8703	0.8891	0.8442	0.8196	0.8334	0.9176	0.5449	0.6281
SMNMF (1)	0.9916	0.9889	0.9955	0.9923	0.9892	0.9733	0.9630	0.9579
SMNMF (2)	0.9843	0.9606	0.9908	0.9780	0.9791	0.9596	0.9401	0.9004
SMNMF (5)	0.9822	0.9475	0.9548	0.9662	0.9486	0.9258	0.9037	0.9399
SMNMF (10)	0.9758	0.9278	0.9615	0.9183	0.9390	0.8940	0.8814	0.9203