Research Article

CDCN: A New NMF-Based Community Detection Method with Community Structures and Node Attributes

Table 3

The performances of different community detection algorithms on nonoverlapping communities measured by the AC and NMI.

MetricDatasetsInfoMapCPMSLPALouvainDEMONSNMFCANSMRNCPLC-DCSCICDECDCN

ACCornell0.20.4460.2390.2660.3770.3710.4460.4150.3170.3480.3540.4490.569
Texas0.2140.4710.5230.2690.4750.4960.5450.4700.5400.3690.4880.4980.641
Washington0.1430.4690.4340.2040.3810.4100.4910.5080.4560.4080.4010.5680.695
Wisconsin0.1520.4710.2620.2230.4300.3860.4710.4710.4220.3540.3960.6450.694
Citeseer0.0530.1780.0900.2380.2080.3090.2120.2110.3140.4520.3270.4740.544
HEP-TH0.0410.1230.0910.1350.1130.1680.1350.1620.1430.1930.1840.2010.215
NMICornell0.1470.050.1380.1090.0510.0610.0450.0610.0840.0810.0730.3110.358
Texas0.1020.0770.0400.0590.0430.0970.0210.090.1150.0680.0970.2520.328
Washington0.1170.0060.1580.0870.0790.0350.0460.1170.0380.1030.0860.3410.406
Wisconsin0.1110.0270.1060.0780.0470.0700.0370.0700.0770.0710.0690.4060.427
Citeseer0.2140.1990.2140.2280.1660.0960.0030.0070.0030.2210.0830.2080.263
HEP-TH0.0350.0920.0840.1120.0930.1370.1140.1230.1140.1520.1450.1790.191