Personalizing information appliances: a multi-agent framework for TV programme recommendations

Abstract

Abstract It has been advocated to develop information appliances to provide ubiquitous Internet information access. After the invention of digital set-top-box, television is expected to become one of the most popular information appliances because of the tremendous digital multi-media programmes it can broadcast. However, as in the World Wide Web, the available TV programmes and their correspondingly electronic information within the increasing digital channels may cause the problem of information overload. Though the audience has more alternatives while choosing programmes, he also has to spend more and more time to read the on-line information about the programme contents or to browse different channels in order to decide what to watch. One way to overcome such a problem is to build intelligent recommender systems to provide personalized information services. By analyzing the information collected from the user, a personalized recommender system is able to reason his personal preferences and then choose the programmes for him. This paper presents a multi-agent framework in which a decision tree-based approach is proposed to learn a user's preferences. The experimental studies concentrate on how to recommend programmes of films and news to a user, and on how the system can adapt to a user's most recent preferences. The results and analysis show the promise of our system.