Effectively Detecting Communities by Adjusting Initial Structure via Cores

Table 3

A quantitative comparison of results generated by ACC and its baselines on networks without ground-truth community structures.

Networks	Measures	ACC	SC	New06	Info	FastQ	PPC	Attractor	LPA
Email	Q	0.466	0.325	0.489	0.535	0.507	0.559	0.464	0.393
	Communities	78	48	7	60	11	16	145	45
CA-GrQc	Q	0.741	—	0.782	0.842	0.812	0.846	—	0.737
	Communities	523	—	394	373	426	449	—	1053
CA-HepTh	Q	0.604	—	0.613	0.742	0.714	0.742	—	0.620
	Communities	961	—	463	459	547	10	—	1775
DBLP	Q	0.672	—	—	—	0.735	0.790	0.592	0.673
	Communities	13212	—	—	—	3090	4108	36043	20874
Amazon	Q	0.737	—	—	—	0.869	0.901	44513	0.786
	Communities	25933	—	—	—	1481	3469	0.668	22278

SC, spectral clustering; New06, Newman2006; Info: infohiermap.