Effectively Detecting Communities by Adjusting Initial Structure via Cores

Table 2

A quantitative comparison of results generated by ACC and its baselines on networks with ground-truth community structures.

Networks	Measures	GT	ACC	SC	New06	Info	FastQ	PPC	Attractor	LPA
Risk Map	Q	0.634	0.634	0.619	0.547	0.634	0.625	0.634	0.462	0.594
	Accuracy	1.000	1.000	0.857	0.762	1.000	0.810	1.000	0.571	0.858
	NMI	1.000	1.000	0.884	0.756	1.000	0.855	1.000	0.754	0.883
	ARI	1.000	1.000	0.599	0.610	1.000	0.709	1.000	0.367	0.753
Dolphins	Q	0.518	0.520	0.379	0.491	0.525	0.491	0.524	0.274	0.514
	Accuracy	1.000	0.984	0.677	0.710	0.887	0.839	0.887	0.565	0.823
	NMI	1.000	0.949	0.592	0.635	0.898	0.733	0.902	0.529	0.792
	ARI	1.000	0.945	0.420	0.496	0.828	0.680	0.828	0.269	0.691
Karate	Q	0.371	0.371	0.313	0.393	0.402	0.381	0.402	0.371	0.365
	Accuracy	1.000	1.000	0.912	0.618	0.824	0.735	0.823	0.971	0.820
	NMI	1.000	1.000	0.646	0.677	0.699	0.692	0.711	0.927	0.690
	ARI	1.000	1.000	0.669	0.512	0.702	0.680	0.702	0.939	0.661
Polbooks	Q	0.415	0.495	0.448	0.467	0.526	0.502	0.419	0.501	0.500
	Accuracy	1.000	0.810	0.495	0.714	0.781	0.800	0.580	0.648	0.648
	NMI	1.000	0.566	0.434	0.520	0.541	0.523	0.395	0.487	0.526
	ARI	1.000	0.681	0.307	0.547	0.646	0.623	0.340	0.485	0.601
Santa Fe	Q	0.739	0.718	0.705	0.708	0.651/0.704	0.749	0.740	0.698	0.674
	Accuracy	1.000	0.847	0.788	0.831	0.636/0.602	0.831	0.771	0.653	0.754
	NMI	1.000	0.886	0.851	0.834	0.764/0.805	0.867	0.880	0.808	0.753
	ARI	1.000	0.827	0.493	0.735	0.582/0.574	0.741	0.760	0.668	0.599
Football	Q	0.601	0.560	0.559	0.492	0.600	0.550	0.553	0.600	0.586
	Accuracy	1.000	0.852	0.835	0.652	0.991	0.548	0.878	0.991	0.924
	NMI	1.000	0.952	0.923	0.762	0.989	0.762	0.887	0.989	0.960
	ARI	1.000	0.853	0.783	0.528	0.982	0.512	0.802	0.982	0.886

GT, ground-truth; SC, spectral clustering; New06, Newman2006; Info: infohiermap.