Collusive shill bidding detection in online auctions using Markov Random Field

Highlights：

• Shill bidding is a strategy employed by a seller who submits fake bids into an online auction to inflate an item’s final price, thereby cheating legitimate bidders.

• Shill bidding detection becomes more difficult when a seller involves two or more bidders to commit shill bidding collaboratively in his/her auction.

• The reason is colluding shill bidders can distribute the work evenly among each other to collectively reduce their chances of being detected.

• This paper presents a collusive shill bidding detection algorithm based on Markov Random Field for identifying colluding shill bidders.

• We implemented the proposed algorithm and applied it on simulated and commercial auction datasets.

• Experimental results on the simulated auction datasets show that the algorithm can potentially detect colluding shill bidders with about 99% detection accuracy.

• Two existing published approaches applied on the simulated auction datasets achieve a detection accuracy of 85% and 88% approximately.

摘要

•Shill bidding is a strategy employed by a seller who submits fake bids into an online auction to inflate an item’s final price, thereby cheating legitimate bidders.•Shill bidding detection becomes more difficult when a seller involves two or more bidders to commit shill bidding collaboratively in his/her auction.•The reason is colluding shill bidders can distribute the work evenly among each other to collectively reduce their chances of being detected.•This paper presents a collusive shill bidding detection algorithm based on Markov Random Field for identifying colluding shill bidders.•We implemented the proposed algorithm and applied it on simulated and commercial auction datasets.•Experimental results on the simulated auction datasets show that the algorithm can potentially detect colluding shill bidders with about 99% detection accuracy.•Two existing published approaches applied on the simulated auction datasets achieve a detection accuracy of 85% and 88% approximately.