Abstract

Information technology is playing an increasingly important role in today's world. Commerce through Internet is not an exception to this phenomenon. Currently the focus in the retailer e-commerce is shifting toward catering to the needs of repeat customers by offering them more personalized services. One of the barriers to such an individualized approach to each customer is the difficulty of collecting information about individual users. This thesis addresses this knowledge acquisition problem. Based on a thorough analysis of different kinds of knowledge acquisition tools and techniques, we propose an architecture that allows the use of a combination of different approaches for knowledge acquisition about users in e-commerce. This architecture is designed to support dynamic adaptation of the user profile to changes in the user interests as well as in the store. The architecture is based on two core concepts, namely dynamic personalization and software agent-support. To reduce the time and effort put in the process of knowledge acquisition by the user and by the knowledge engineer software agents in the proposed architecture assist in different aspects of the process, such as profile initialization, processing of results discovered by the web mining, making changes to user profile and tracking their effects, and in trust related issues and interaction with other agents and systems. A proof of concept prototype has been implemented to demonstrate the feasibility of the architecture