Detecting suspicious behavior using a graph-based approach

The ability to discover illicit behaviour in complex, heterogeneous data is a daunting problem. In the VAST 2014 competition, one of the challenges involves identifying for local law enforcement which employees are involved and where they should be concentrating their efforts. One approach to handling this problem is a graph-based approach. In this paper, we present a graph-based anomaly detection approach for discovering suspicious employees and geographic locations.