Aplicação de técnicas de processamento de linguagem natural na análise de feedbacks acadêmicos

Abstract

Students at UTFPR Ponta Grossa campus are invited to participate in the Student Evaluation of Teaching (ADPD) at the end of each semester. The ADPD is a brief evaluation containing both quantitative and qualitative questions, providing professors and administrators with greater insight to support future decision-making based on students' classroom experiences, as well as the opinions and suggestions collected through the survey. Therefore, this work aimed to develop a browser extension capable of performing sentiment analysis on textual feedback provided by students to their instructors, assisting in the identification of patterns and contributing to educational evaluation processes and institutional decision-making. The solution was developed based on artificial intelligence techniques, specifically approaches involving Natural Language Processing (NLP) and supervised machine learning. The system was designed to operate entirely locally on the client side, enabling its use directly within the browser without the need for external servers. This approach reduces infrastructure complexity, increases the portability of the solution, and contributes to the privacy of the processed data. The system validation is carried out through functional testing and observation of the tool's behavior, analyzing its ability to correctly classify feedback within the proposed educational context. Finally, it is concluded that the developed solution demonstrates potential for application in educational environments and may serve as a foundation for future improvements and adaptations to other domains involving automatic text classification. The code used to create the extension was structured to be scalable and easy to modify, increasing its potential for adoption across different educational platforms