Many existing solutions for the automatic assessment of open-ended questions predominantly rely on machine learning models, primarily focusing on aspects such as writing style and assigning a final score. However, these solutions often overlook the crucial factor of feedback content relevance, specifically, how well the response aligns with the content of the original question. This research introduces a novel approach aimed at enhancing the rapid feedback essential for this type of assessment. The approach involves identifying individual cognitive deficiencies among students and providing guidance for their remediation. The primary objective is to seamlessly integrate pedagogical guidelines founded on competencies and skills by leveraging an educational recommendation system. This system incorporates the concepts of ontology learning, ontology alignment algorithms, action recommendation algorithms tailored to each student's unique needs. As the main outcomes, a case study is presented, illustrating each step of the system.