Abstract. This paper designs an electronic commerce system that integrates conventional electronic commerce services with contemporary WWW advantages, such as comprehensive coverage and agents for information search and selection. We use a user-centered approach and apply data mining techniques in the design of agents for information search and selection. There are various agents in this electronic commerce system to perform different functions. Among them, SiteHelper is a unique agent in our system compared to existing electronic commerce systems. It acts as a housekeeper for the system and as a helper for the users to find relevant information. In order to assist the users in finding relevant information at the centralized location (with Web links to the global Web), SiteHelper interactively and incrementally learns about each user’s areas of interest and aids them accordingly, by deploying data mining techniques with incremental learning facilities as its learning and inference...