An agent-based platform for online auctions

This paper proposes a web-based agent platform for E-commerce which allows humans and software agents to perform automatic auctions over the Internet. Internet-based auction is a profitable, exciting and dynamic part of E-commerce. However, the lack of standard on negotiation protocol between agents and an auctioneer makes full automation of E-commerce infeasible. Hence, we design a complete architecture and a set of negotiation protocols based on advanced agent technologies. Moreover, wr evaluated the negotiation protocol using a Markov chain model. Our experimental results show that (1) thc Markov chain can model independently simulated bidder's behaviour accurately, (2) a wider range of bidder's behaviour can be simulated with agents having partial knowledge of other bidders' willingness to bid and the closing time. In summary, our agent-based platform can incorporate realistic scenarios for simulation in online auction and other E-commerce applications.