Combining Data Mining Technique and Users' Relevance Opinion to Build an Efficient Recommender System

The huge amount of information on the Internet creates a problem for the users – information overload. For this reason, finding the worthwhile information is becoming a challenge. To aid users a new approach based on Recommender System. This type of system applies information filtering in order to recommend items to a user based on the user's profile and historical consumption. Recommender Systems present some difficulties: (i) user overspecialization and (ii) the new user problem. The main contribution of this paper is the description of a Framework to discover new knowledge based on a data mining technique and user's relevance opinion. This knowledge is represented as a set of rules in a knowledge base, which has been used to address the difficulties cited before and to help in the information filtering process. This paper reports on work, which is part of the W-RECMAS (a Recommender System to Web based on Multi-Agent System for academic paper recommendation) project.