Efficiently identifying critical nodes in large complex networks

Computational Social Networks

Table 4 Comparison of the CNDP objective value after removing 10% and 20% of vertices from each network in Table 1

Problem	K	SEQ	Degree	PageRank	Authority	MIS
Comnat	2313	58,796,393	103,398,683	87,630,163	126,804,602	NA
	4627	83,686	90,610	92,242	7,399,785	NA
Ego	404	2,717,347	5,339,614	3,816,109	6,320,816	2,192,636
	808	1,848,740	2,070,535	2,886,709	3,438,031	903,441
Flight	294	322,527	484,331	467,962	1,014,305	77,777
	588	1,457	1,698	1,715	1,567	2,626
Powergrid	494	22,182	51,508	212,369	56,815	25,253
	988	3,639	4,580	14,744	3,771	5,378
Relativity	524	224,010	1,628,337	302,309	3,382,195	23,620
	1,048	4,089	4,896	9,023	6,390	6,163
Oclinks	190	637,936	785,662	758,328	835,297	746,085
	380	218,215	258,277	246,876	306,289	402,824

SEQ-based results are those obtained by the proposed algorithm. The MIS-based approach was unable to arrive at solutions within 40 h for the conmat networks.