Recommendation Based Discovery of Dynamic Virtual Communities

Recommendation systems are becoming increasingly popular in various large-scale web-based applications (such as infomediaries, emarketplaces, knowledge portals) since they enable users and customers to better fulfil their information or product search needs according to their interests. Techniques used to form the recommendations include algorithms from the area of machine learning and vote theory. We introduce an alternative way of building user profiles based on explicit and implicit ratings as well as profile-based recommendation and user-matching algorithms. The paper advocates the use of such algorithms as a tool to automate the discovery and creation of dynamic, virtual communities. Furthermore, rating prediction algorithms are proposed as a means of enhancing the interaction between users and a recommendation system. As an application of the proposed techniques, we present MRS, a web-based information system that provides personalized recommendations for cinema movies.

• Type: Conference
• Year: 2003
• Where: CAISE
• Authors: Manos Papagelis, Dimitris Plexousakis
• Keywords: CAISE 2003, Large-scale Web-based Applications, Recommendation System, Recommendations Include Algorithms