Table 1.

The table shows the following features of 10 covert networks: network size (N), the number of edges (E), the graph density (D), the average network modularity according to Louvain method (Q), the best known topological targeted attack, the network robustness to that attack (R, see Eq. 2) and the degree assortativity (A). BFP2013 is the Brazilian Federal Police criminal intelligence network¹³.

Network	N	E	D	Q	Att	R	A
BFP2013	9,887	19,744	4E-4	0.96	HBA	0.007	0.02
ISIS	56	30	1E-2	0.50	CI	0.020	−0.39
CAVIAR	110	3,546	3E-1	0.65	CI	0.040	−0.28
DRUG	293	337	4E-3	0.75	CI	0.040	0.06
IRA	83	184	3E-2	0.75	CI	0.060	−0.04
FIFA	450	10,859	5E-2	0.74	CI	0.060	0.24
GANGS	67	228	5E-2	0.56	CI	0.070	−0.35
9/11	61	199	5E-2	0.64	CI	0.130	−0.16
Full	10,407	842,247	8E-3	0.20	HiDA	0.055	−0.26
Strong	766	112,460	0.19	0.16	HBsA	0.347	−0.21

The ISIS dataset consists of the relationships between members of the terrorist group supplied according to the BBC⁴². CAVIAR is the dataset from an investigation of drug trafficking operating out of Canada³⁵. DRUG stands for the network of drug users in Hartford, USA⁴³. IRA is the network data on active Irish Republican Army members between 1970 and 1998⁴⁴. FIFA is the network of covert elements from a recent scandal involving the Fédération Internationale de Football Association⁴⁵. The GANGS dataset describes Italian gang members⁴⁶. 9/11 is the network data of the bombing of the World Trade Centres in 2011 constructed from news reports⁴⁷. The network studied in this paper is split “Full” for the whole graph and “Strong” for the strongly connected component.