Advances in Personalised Recommendation of Learning Objects Based on the Set Covering Problem Using Ontology

Loop-based intelligent tutoring systems (ITSs) support the learning process using a step-by-step problem-solving approach. A limitation of ITSs is that few contents are compatible with this approach. On the other hand, recommendation systems can recommend different types of content but ignore the fine-grained concepts typical of the step-by-step approach. This work contributes to the solution of this state-of-the-art challenge by proposing an approach for the recommendation of learning objects from different areas of knowledge, considering the refined concepts of ITSs. To deal with this challenge, we formulate the learning object recommendation problem as the set covering problem that belongs to the NP-hard class problems. An exact algorithm and a greedy heuristic were properly adapted, resulting in a promising approach to solve these problems, as shown by the results. This resulted in more personalised content for students using collaborative filtering and an ontology that models their knowledge, learning styles and search parameters.