Module detection in complex networks using integer optimisation

Table 1 Computational results comparing the performance of modularity optimisation methodologies across several network examples.

Networks			iMod			EB	EIG	C₃/C₄	EO	SA	QCUT	Greedy
Name	N	L	Median Q	Best Q	M	Q
Zachary	34	78	0.420	0.420	4	0.401	0.419	0.417	0.419		0.420	0.419
Dolphin	62	159	0.529	0.529	5	0.520					0.518	0.519
Les Miserables	77	254	0.560	0.560	6	0.540					0.560	0.556
P53	104	226	0.535	0.535	7						0.522	0.531
Jazz	198	2742	0.445	0.445	4	0.405	0.442	0.441	0.445		0.445	0.443
E. coli	418	519	0.780	0.781	19		0.766			0.752	0.776	0.779
S. cerevisiae	688	1079	0.768	0.775	25		0.759			0.740	0.766	0.764
C. elegans	453	2025	0.451	0.453	9	0.403	0.435	0.422	0.434		0.433	0.441
Email	1133	5451	0.575	0.580	9	0.532	0.572	0.567	0.574		0.576	0.543

Best modularity achieved across all methodologies and network examples is denoted in bold.
References: EB [25], EIG [9], C₃/C₄ [51] EO [32], SA [33], QCUT [34], Greedy [52]

ISSN: 1748-7188