Social Network Analysis for Fraud Detection

• Activity Number: 21
• Type: Topic Contributed
• Date/Time: Sunday, August 1, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): John Clare Brocklebank*+
• Companies: SAS Institute
• Keywords: Social Network Analysis ; Financial Crimes ; Fraud Detection ; Anomaly Detection ; Predictive Models ; Business Rules

Abstract

As stocks and housing prices fall and consumer confidence sinks in today's economic turmoil, one area of growth has been crime and fraud. Banks, insurance companies, and government entities are seeing an increase in both the number and sophistication of fraudulent activities.

To fight fraud effectively, organizations must improve the monitoring of customer behavior across multiple accounts, systems, and agencies. They must develop a framework of components that support fraud detection, alert generation, and case management. Using a hybrid approach, the framework can include industry-specific business rules, anomaly detection, predictive models, and social network analysis. It can offer both top-down and bottom-up functionality for making hidden and risky networks visible to investigators.

This methodology will be illustrated with examples from banking and insurance.