Analysis of Student Sentiment Toward the Use of ChatGPT for College Assignments Using the Naïve Bayes Method

Abstract

Advances in generative artificial intelligence technology have driven an increase in the use of ChatGPT for various academic activities, including the completion of college assignments. This study aims to analyze student sentiment toward the use of ChatGPT for college assignments using the Naïve Bayes method. The study employs an empirical, machine learning-based experimental approach using a dataset obtained from Kaggle consisting of 1,153 student comments. The research stages include data preprocessing, feature extraction using Term Frequency–Inverse Document Frequency (TF-IDF), training the Naïve Bayes model, and evaluating performance through a confusion matrix and classification accuracy. The preprocessing results yielded 1,125 clean data points suitable for analysis. The sentiment distribution showed a dominance of negative sentiment in 1,003 data points (89.16%), while positive sentiment accounted for 122 data points (10.84%). Model testing yielded an accuracy rate of 90.67%, indicating that the Naïve Bayes algorithm is capable of effectively identifying sentiment patterns in student textual data. The research findings indicate that although ChatGPT offers benefits in improving task completion efficiency, most students still have concerns regarding academic dependency, academic integrity, and a decline in critical thinking skills. The research results contribute to our understanding of students’ perceptions regarding the use of generative artificial intelligence in higher education settings.