Role Recognition of Illegal Online Gambling Participants Using Monetary Transaction Data

摘要

Online gambling has become a substantial global industry during the past two decades. However, it is explicitly prohibited or restricted by most countries in the world due to social problems caused by it. This results in rapid expansion of the illegal online gambling (IOG) market where players profits are under little protection. To fight against IOG, this paper addresses the IOG participant-role recognition (PRR) problem by learning a supervised classifier with monetary transaction data. We propose two sets of features, i.e., transaction statistical features and network structural features, to effectively represent participants. Based on the feature representation, we adopt an ensemble learning strategy in the training phase of a PRR classifier to reduce the impact of unbalanced data. Results of experiments performed on real-world IOG case data demonstrate the feasibility and validity of the proposed approach. The proposed approach could help investigators in a law enforcement agency find the key members of an IOG organization quickly and destroy the ecosystem efficiently.