Data science research aims to unravel crime networks

17 August 2021

An international research collaboration involving the University of Derby is examining how social network analysis could disrupt some of the world’s most dangerous criminal gangs.

With organised criminal groups doing their best to frustrate law enforcement agencies, this research aims to show that information technology and data analysis may be used to effectively unravel their structure, and to identify the links between the key figures in their networks.

Student Lucia Cavallaro, currently in the third-year of her PhD at the Data Science Research Centre at Derby, has been investigating the topic of Network Features in Complex Networks, inspired by data from criminal investigations, with potential impact on active crime groups.

Working with her supervisor Dr Ovidiu Bagdasar, Associate Professor in Mathematics at the University, and in collaboration with professors and fellow PhD students from the Universities of Messina and Palermo in Sicily (Annamaria Ficara and Francesco Curreri) and Bolzano in Italy, Lucia has used data extracted from Italian court proceedings (in particular, the calls and meetings between members of two Mafia families), and studied the resulting networks by data analysis.

Lucia said: “By developing new algorithms to analyse the data, we identified that criminals tend to minimise interception risks, preferring to spread their messages within the clan with a balanced number of intermediaries. It is also clear that by identifying and removing key individuals, or ‘nodes’, which are shown on graphs of the networks created by algorithms, we can help the police to effectively disrupt these groups. We also found that one particular model – Barabasi-Albert – provides the most accurate means of artificially replicating the properties of criminal networks.”

A key finding of the project was that the investigations of Law Enforcement Agencies (LEAs) remain effective even when a considerable percentage of the calls between gang members are missed, confirming that it is more important to know who the gang members are, rather than know all they do.

Lucia’s work complements two studies led by Antonio Liotta, currently Professor of Data Science at the Free University of Bolzano-Bozen, and Head of the Data Science Research Centre at Derby until 2019, as well as that of colleagues in Italy and China.

A concept known as Betweeness Centrality, a measure of the most effective communication path in a network, enabled the research teams to understand who played decisive roles in the sharing of information within the group, such as ordering a crime, or arranging an illicit deal. Once these nodes have been identified, targeted arrests or other police action can take place in order to weaken the entire network.

Professor Liotta said: “Understanding which nodes are the fundamental ones on the knowledge graph can maximise the ability to reduce any communication across the network, even if the police do not succeed in arresting the crime boss because he does not show up in the network except through trusted intermediaries. If you isolate key elements, you give law enforcement time to go after the boss. This way we can minimise the boss’ ability to restore his criminal network.”

The researchers have also developed algorithms generating a “synthetic” criminal network, which can be tailored to simulate any organisation, even if the information about that group is very limited – particularly given the confidentiality of police investigations.

The research has generated interest among academics in the USA, who are studying Mafia groups in America, while a second paper produced by Professor Liotta and his colleagues, has examined groups operating across the world, from Canada to the Philippines.

Work is now underway to anonymise sensitive data, which the academics hope will lead to more direct collaborations between them and law enforcement agencies in future.