Real-Time Model Checking for Shill Detection in Live Online Auctions

Online auctions are vulnerable to shill bidders, who intend to artificially raise bidding prices, causing winning bidders to pay more than they should pay for auctioned items. Detection of such fraudulent behaviors is very difficult, especially when an auction is in progress, or “live”. This paper focuses on a formal technique to detect shilling behaviors in live online auctions. We define a monitoring agent that can continuously watch for abnormal bidding behaviors of a monitored bidder. To make the detection process efficient, we introduce a dynamic auction model (DAM), and use real-time model checking techniques to verify shilling behaviors specified in linear temporal logic (LTL). Finally, we present an algorithm for real-time shill detection, and use a case study to demonstrate the efficiency and effectiveness of our approach.