Identifying Important Nodes in Weighted Covert Networks Using Generalized Centrality Measures

Abstract

For investigators working on criminal covert networks, identification of key actor(s) in the network is a major objective. Taking out key nodes will decrease the ability of the criminal network to function normally. Traditionally, the node centrality measurements have relied solely on the number of edges incident to nodes but not on the weights of those edges. However, in some generalizations for centrality measures for weighted networks, the focus shifts solely to the weights of the links and they don't account for the number of ties which was the central idea in the original centrality measures. Hence, answering which nodes are most central in a network with weighted relations depends on what imporantce is given the weights of the incident edges in comparison to the number of those edges. Opsahl et al. propose a generalized method for controlling the relative importance between the number of incident ties (nodal degree) versus the total weight of those ties (nodal strength). Research in TNA has largely focused on un-weighted ties, whereas richer and more sophisticated models of covert networks are needed to give precise and more realistic knowledge about such networks. Moreover, the existing implementation of node centrality algorithms in TNA tools don't support networks having weighted/values relations among nodes. New implementations of node centrality algorithms for weighted networks based on the generalized approach have been developed in Crime Fighter Assistant tool and are evaluated with known network dataset of the 9/11 incident.