An Approach to Ranking Participants Based on Relationship Network in E-commerce

Abstract

In recently years, many people easily access Internet auctions in e-commerce trading. At the same time, network structures like the WWW have become huge and are analyzed on a grand scale. In Internet auctions, users face the problem of really knowing the credit and trustworthiness of participants, and the simple rating mechanism widely used in Internet auctions fails to represent this accurately. This paper proposes participant ranking methods based on relationships in Internet auctions. Our algorithm called "Auction Network Trust (ANT)" employs HITS's techniques and Internet auction data. At this stage, we successfully implemented a crawler for Internet auction sites and compared our algorithm to a reputation value of Internet auctions with several approaches such as user rankings. Furthermore, our work possesses a network analyzing system on a larger trading network that predicts which buyers and sellers are active and demonstrate better behaviors. Our experiments show many behaviors in the Internet auctions and that ANT presents different scores from HITS on the WWW.