Identification of shill bidding for online auctions using anomaly detection

Abstract

The paper presents a novel method of shill biding frauds detection in online auctions. The main idea behind the method is a reputation system using anomaly detection techniques. The system focuses on cases where the final price can be inflated by interference of persons who are colluding with the seller. The main aim of the work was to support users of online auctions systems by mechanisms which would be able to detect this type of frauds. The proposed method of shill bidding identification has been implemented using statistical analysis software and data derived from the test bed provided by one of the leading online auction houses. The other aim of the research was to assess whether the proposed solution is better than previous approaches described in the literature and how well the systems are able to detect real frauds. The presented system has been validated using some experimental data obtained from real world auction systems and specially generated with application of domain specific tools. Study confirmed that proposed system was able to detect most frauds related to the artificial price inflation.