This paper is part of a doctoral thesis that aims to propose an evaluation model, for later application, using Educational Data Mining techniques to analyze the responses of students obtained during an Institutional Teaching Evaluation. Therefore, the authors propose an Institutional Teaching Evaluation model that applies, among others, the Sentiment Analysis to identify which teaching practices are positive or negative from the perspective of students from a Higher Education Institution. Differently than the current Institutional Teaching Evaluation models, which are founded on the evaluators' assumptions, this model responds to the need of better pedagogical practices through data mining, finding new categories of analysis in the discourse of a group of students to contribute to the conception of a more effective Teaching Evaluation and to create awareness about teaching practices.