Handy broker: An intelligent product-brokering agent for m-commerce applications with user preference tracking

Abstract

One of the potential applications for agent-based systems is m-commerce. A lot of research has been done on making such systems intelligent to personalize their services for users. In most systems, user-supplied keywords are generally used to help generate profiles for users. In this paper, an evolutionary ontology-based product-brokering agent has been designed for m-commerce applications. It uses an evaluation function to represent a user's preference instead of the usual keyword-based profile. By using genetic algorithms, the agent tracks the user's preferences for a particular product by tuning some parameters inside its evaluation function. A prototype called 'Handy Broker' has been implemented in Java and the results obtained from our experiments look promising for m-commerce use. © 2002 Published by Elsevier Science B.V.