E-learning is an electronic teaching model that can be adapted to learning styles of every student. In this context, the lack of learning objects (LOs) that can be recommended to students is an outstanding problem. Creating educational content is often a time-consuming task, which has a high cost in money. To solve this problem, we create an approach that uses Wikipedia content to create new educational resources and an ontology for modeling metadata and students. The second challenge is the recommendation of LOs in accordance with user search parameters and the student profile. The Problem of Recommendation of Learning Objects is formalized in this work. This problem is based on concept coverage, the goal is to recommend the LOs that cover the concepts that the student is expected to learn. It is solved through inference rules and a genetic algorithm, taking into account also the search parameters of the user and the learning styles of learners. The tests performed on a prototype ensure the feasibility of this approach for the creation and recommendation of LOs.