Recommender System for Learning Objects based in the Results of Learner Choices in E-learning Systems - Application in the MERLOT Platform

Abstract

In this paper, we briefly review the recommendation approach of learning objects (LOs) that uses the result of the choices made by the learner user during learning (“what to learn”, “how to learn”, etc.) as a source of information. This LO recommendation approach is an implementation of collaborative filtering based on an instance-based machine learning method. The goals of this paper are: to present how to apply this LO recommendation approach in an e-learning platform used for non-formal learning – the MERLOT platform, and to perform an experimental evaluation on the MERLOT’s dataset. The evaluation showed that the LO recommendation approach presents higher prediction accuracy than the baseline approaches.