In the educational context, recommendation systems play a crucial role in personalizing learning experiences by considering cognitive, pedagogical, and affective dimensions. Their main objective is to promote learner engagement, motivation, and performance. However, these systems are often subject to demographic, linguistic, or cultural biases that can compromise equity and accessibility. This article proposes the FairEduSystem model, based on a conceptual approach and techniques proven in the literature, such as reweighting, diversified reclassification, and an equity-sensitive cost function. This model aims to provide personalized educational recommendations while ensuring fairness, diversity, and transparency, and integrating ethical considerations from the pre-processing stage of the data. Future implementation will allow for an assessment of its actual effectiveness and impact on reducing bias and enhancing learner outcomes.