GOTCHA! Network-based Fraud Detection for Social Security Fraud

V. Van Vlasselaer, T. Eliassi-Rad, L. Akoglu, M. Snoeck, B. Baesens
Management Science (INFORMS)
pp. 1-21, July 2016.
July 14, 2016

Abstract

We study the impact  of network information for social security fraud detection. In a social  security system, companies have to pay taxes to the government. This study  aims to identify those companies that intentionally go bankrupt to avoid  contributing their taxes. We link companies to each other through their  shared resources, because some resources are the instigators of fraud. We  introduce GOTCHA!, a new approach to define and extract features from a  time-weighted network and to exploit and integrate network-based and  intrinsic features in fraud detection. The GOTCHA! propagation algorithm  diffuses fraud through the network, labeling the unknown and anticipating  future fraud while simultaneously decaying the importance of past fraud. We  find that domain-driven network variables have a significant impact on  detecting past and future frauds and improve the baseline by detecting up to  55% additional fraudsters over time.

Related publications