An approach to recommender system applying usage mining to predict users’ interests

This paper presents an approach to predict users´ interests in Web systems based on a collaborative filtering method, which takes into account implicit information about the users´ navigation. To test and evaluate the proposed approach, a particular case study, with values established by a Web based educational systems (WbE-S) domain specialist, was conducted. The results were compared using a cross-validation method. The comparison shows that the proposed approach, based on real users relationship with the system, produces good predictions for a recommender system.