Efficient E-Commerce Agent Design Based on Clustering eBay Data

In this paper we present the use of data mining techniques on data derived from eBay auctions in order to design a decision making mechanism for autonomous agents. These agents operate as parallel threads that concurrently receive data from multiple online auctions, where a specific item is traded. The agent objective is to select the auction, in which the traded item will be sold at the lowest price. Trading agents are armed with a realtime decision-making mechanism, adaptable to changes occurring in the auctions environment. The paper shows how a particular data mining technique can be incorporated into the design of trading agents and provides the theoretical basis of the agent decision mechanism.