Discovering user interests from Web browsing behavior: an application to Internet news services

Discovering user interests is a very important task for providing personalized services in electronic commerce. A popular approach is to develop customer profiles from their browsing behavior. In this paper, we present an approach that analyzes the browsing content and time to determine user interests. An empirical study using actual news provided by the China Times shows that the proposed system outperforms the traditional headline news compiled by the news editor in both objective performance indices and customer satisfaction.