Bipartite Network Model for Inferring Hidden Ties in Crime Data

TL;DR

This paper introduces a bipartite network model to uncover hidden relationships in crime data, aiding in understanding and disrupting organized criminal networks through structural and community analysis.

Contribution

We propose a novel bipartite network approach to infer unseen ties in crime data and validate it with real-world case studies on pharmaceutical and online forum crimes.

Findings

Effective identification of hidden criminal ties

Community structures reveal operational organization

Potential to improve disruption strategies

Abstract

Certain crimes are hardly committed by individuals but carefully organised by group of associates and affiliates loosely connected to each other with a single or small group of individuals coordinating the overall actions. A common starting point in understanding the structural organisation of criminal groups is to identify the criminals and their associates. Situations arise in many criminal datasets where there is no direct connection among the criminals. In this paper, we investigate ties and community structure in crime data in order to understand the operations of both traditional and cyber criminals, as well as to predict the existence of organised criminal networks. Our contributions are twofold: we propose a bipartite network model for inferring hidden ties between actors who initiated an illegal interaction and objects affected by the interaction, we then validate the method in two case studies on pharmaceutical crime and underground forum data using standard network algorithms for structural and community analysis. The vertex level metrics and community analysis results obtained indicate the significance of our work in understanding the operations and structure of organised criminal networks which were not immediately obvious in the data. Identifying these groups and mapping their relationship to one another is essential in making more effective disruption strategies in the future.