A Transactional Relationship Visualization System in Internet Auctions

As Internet auctions have increased, so too has auction fraud. This paper describes the design of a system that supports fraud detection and market prediction by visualizing transaction networks on Internet auctions. Our system employs link mining techniques and user information. At this stage, we successfully showed visualized trading networks by extracting trading histories from an auction site. Using a visualized graph, the system shows suspiciousness with user ID information. Also, the system presents trading relationships as a network structure in various viewpoints. Furthermore, the system possesses an explanation function on a visualized trading network that predicts which buyers and sellers are active and did better behaviors. Our preliminary experiment demonstrates that the graph presentation function is scalable enough against the number of sellers and buyers.