Since malicious bidding behaviors are changed frequently, we need a profiling mechanism to profile malicious bidding behaviors for detecting shill bidder. In this study, we proposed a malicious bidding prevention model (MBP). This model can profile malicious bidding behavior for detecting potential shill bidder and reduce computational cost in detecting. There are three modules in this model, namely, profiled malicious bidder behavior module (PMBB), shilling behavior detection module (SBD), and reputation module (RM). We applied self-organizing map (SOM) algorithm to build up PMBB and SBD. PMBB can profile malicious bidder behavior by using SOM algorithm. SBD detect malicious bidder based on similarity comparison between malicious profile and user behaviors by using SOM algorithm. Finally, RM will mark the malicious reputation for prior detected user. Via the experiment results, the proposed model is effectively capable to detect shilling bidding behaviors and reduce the computational cost and time. Moreover, the auctioneer can force shill seller become normal seller or leave the auction website by punishing him/her.
