Desarrollo de un Sistema Recomendador para la Asignación de Docentes para asignaturas en la Universidad de Cuenca

Abstract

The proper allocation of faculty to courses in higher education represents a key challenge in academic talent management, as it directly affects the quality of the teaching–learning process. Currently, this procedure is often carried out manually, introducing limitations such as subjectivity, lack of standardization, and high administrative workload, which frequently lead to suboptimal assignments. To address this issue, the present study proposes and va- lidates a recommender system designed to optimize the faculty assignment process. The system combines sentiment analysis, through transformer-based language models adapted to the local context, with mathematical models that enable precise alignment between faculty competencies and the specific academic requirements of each course. Faculty profiles were generated by integrating historical evaluations, automatically categorized student feedback, and competencies defined by the institutional Faculty Competency Pentagon. In addition, dynamic weights were incorporated to adjust the relevance of factors according to the aca- demic term, ensuring consistency with institutional expectations. The results demonstrate that the system met the proposed objectives, producing assignments consistent with the cri- teria of program directors in the faculty of engineering. A pilot evaluation with three directors revealed high acceptance and positive assessments of the proposed profiles and assign- ments. Although the limitations of the pilot study prevent broad generalization, the feedback suggests significant potential for application in similar contexts. Overall, the system reduces operational workload and positions itself as a strategic tool with scalability and applicability to other educational environments.