Metodología para la Identificación de Buenas Prácticas Docentes a Partir de Comentarios Estudiantiles Utilizando Análisis de Sentimientos y Modelado Temático

Abstract

This study proposes an empirical methodology to identify the most valued dimensions of teaching practices in higher education through a mixed-method approach that integrates qualitative and quantitative analyses. Over 89,000 student comments collected during teacher evaluations at the University of Cuenca (2024–2025) were analyzed. Advanced natural language processing techniques, specifically sentiment analysis and Structural Topic Modeling (STM), were used to process an extensive textual corpus. The analysis focused on positive comments aimed at teachers recognized through a prior comprehensive evaluation, aiming to uncover recurring patterns in student perceptions. Four key dimensions of student evaluation emerged from the analysis: disciplinary knowledge, communication management, socio-emotional skills, and teaching methodology. These dimensions reflect perceived teaching strengths and exhibit significant variability across disciplinary contexts and academic levels. For example, disciplinary knowledge is prioritized in technical areas, while socio-emotional skills are more valued in social sciences and health disciplines. These findings underscore that student evaluations are an essential source of qualitative information, overcoming limitations of traditional numeric metrics by offering deeper, contextual insights. Ultimately, the proposed methodology provides higher education institutions with a practical tool to identify specific faculty strengths and design evidence-based, personalized professional development programs, thus fostering continuous and contextualized improvements in educational quality.