Conference of the International Federation of Operational Research Societies (IFORS 2014) edition:20 location:Barcelona (Spain) date:13-18 July 2014
Given a bipartite graph consisting of people and activities, can we find groups of persons frequently involved in the same activities? Moreover, given that some activities are fraudulent, are we able to identify those groups of persons that are frequently involved in the same crimes. This study concerns the detection of fraudulent clique structures in a social security context. Using row- and column reordering techniques and simultaneously spreading fraud through the network, our approach successfully succeeds to find cross-associations, as well as to assign a fraud score to each clique.