Comparative analysis of similarity metrics for the collaborative recommendation of learning objects

Collaborative filtering based recommendation systems are based on the premise that if a user looks like another (similar) and that one liked an item, this one will like it too. The collaborative recommendations are made every day in different domains, education is not alien to it because everyday students have access to more educational resources and collaborative recommendations help find those who help in their learning process. One of the difficulties presented in implementing these systems is to determine the best metric of similarity among users among all existing to find a greater amount of similarities to the target user of the recommendation. Therefore, in this paper, we propose to perform a comparative analysis of similarity metrics for the recommendation of learning objects. Tests were conducted with university students and it was found that the overlap coefficient and the distance of the cosine, give better results when making a collaborative recommendation.