Social network analysis approaches to study crime

Social Network Analysis (SNA) studies groups of individuals and can be applied in a lot of areas such us organizational studies, psychology, economics, information science and criminology. One of the most important results of SNA has been the definition of a set of centrality measures (e.g., degree, closeness, betweenness, or clustering coefficient) which can be used to identify the most influential people with respect to their network of relationships. The main problem with computing centrality metrics on social networks is the typical big size of the data. From the computational point of view, SNA represents social networks as graphs composed of a set of nodes connected by another set of edges on which the metrics of interest are computed. To overcome the problem of big data, some computationally-light alternatives to the standard measures, such as Game of Thieves or WERW-Kpath, can be studied. In this regard, one of my main research activities was to analyze the correlation among standard and nonstandard centrality measures on network models and real-world networks. The centrality metrics can greatly contribute to intelligence and criminal investigations allowing to identify, within a covert network, the most central members in terms of connections or information flow. Covert networks are terrorist or criminal networks which are built from the criminal relationships among members of criminal organizations. One of the most renowned criminal organizations is the Sicilian Mafia. The focal point of my research work was the creation of two real-world criminal networks from the judicial documents of an anti-mafia operation called Montagna conducted by a specialized anti-mafia police unit of the Italian Carabinieri in Messina (i.e., the third largest city on the island of Sicily). One network includes meetings and the other one records telephone calls among suspected criminals of two Sicilian Mafia families. This dataset is unique and it might represent a valuable resource for better understanding complex criminal phenomena from a quantitative standpoint. Different SNA approaches have been used on these Montagna networks to describe their structure and functioning, to predict missing links, to identify leaders or to evaluate police interventions aimed at dismantling and disrupting the networks. Graph distances have been used to find a network model able to properly mime the structure of a Mafia network and to quantify the impact of incomplete data not only on Mafia networks such as the Montagna ones but also on terrorist and street gangs networks. The two simple Montagna networks have been finally used to build a multilayer network trying to obtain a more nuanced understanding of the network structure and of the strategic position of nodes in the network.