|Author's Email Address
||This thesis had been viewed 5559 times. Download 349 times.|
|Type of Document
||Applying Clustering to Analyze Bidding Behaviors and Shill Bidding in Online Auction|
|Date of Defense
||In recent years, trading volume of online auction has been climbing steadily. However, the anonymous scheme hides confirm bidder’s identity and bidding history. This may lead to frauds, such as shill bidders participate in the process with the no intention to win but to raise competitive prices.|
We reviewed existing literature to find 7 bidder’s behavior variables that may be used to identify potential shill bidders: (1) frequency of a bidder participated in the same seller, (2) number of bids, (3) number of winning bids, (4) inter-bid time, (5) inter-bid increment, (6) timing of the first bid, (7) timing of the last bid. We used anonymous data to investigate whether shill bidders can be identified by their bidding behavior.
In this research, we have developed a clustering-based approach that uses the increment of reputation and 7 behavior variables of the bidder to determine the probability of shill bidding. A dataset that includes both anonymous bidders and known bidders is used to evaluate the method. Five clusters have been identified from the anonymous data. The likelihood of shill bidding in each cluster was assessed. The data subset of winning bidders was used to evaluate the accuracy of the clustering model. The result indicates that our clustering-based approach can effectively assess the probability of a shill bidder from their bidding behavior. The contribution of this research allows shill bidders to be identified in the bidding process.
||Chiu, Chao-Min - chair|
Ho, Shu-Chun - co-chair
Liang, Ting-Peng - advisor
Indicate in-campus at 1 year and off-campus access at 1 year.|
|Date of Submission