Abstract
Internet auctions are a clear example of how information and networking technologies have transformed traditional trading processes limited to specific events and products, into pervasive and thriving electronic business models enabled by the Internet; they have increased the variety of goods and services that can be traded using auction mechanisms and reduced transaction costs for both buyers and sellers. Nevertheless, due to the characteristics of electronic transactions, it is difficult to distinguish genuine sources of information from fraudulent ones, increasing the vulnerability of participants to be victims of fraudsters. Since Internet auction fraud has become the most frequently reported form of Internet scams, auction websites are trying to provide protection to their customers implementing new services that threaten the low cost structure of this business model and may as well increase the complexity of transactions over time. To counteract the effects of fraud and the threat that it poses to the future of auction websites, specifically the consumer-to-consumer type, this study provides a data mining approach to fraud detection based on variables derived from auctioneers’ transaction history. Using the historical data recorded on Yahoo! Auction site for the transactions of both fraudsters and legitimate auctioneers, our proposed fraud detection model outperforms similar methods aimed at identifying potential fraudsters among participants of Internet auctions.