Evaluating Student Feedback: A Comparative Study of Traditional ML and DistilBERT Model
Tushar Kasana, Baljinder Kaur · 2025
Student reviews are of high importance in today’s digital world to check how good are the educational institutes. This paper analyses student reviews of Indian engineering colleges using Aspect-Based Sentiment Analysis (ABSA). It focuses on important areas like job placements, school facilities, and teachers etc. The paper compares how well traditional machine learning models-Naive Bayes, Support Vector Machines (SVM), and Logistic Regression-work with a newer model called DistilBERT-based ABSA model. The results show where feelings are spread out, giving a sense of where schools may do well or need to get better. In addition to the standard models, a web based interface was developed so users may work with the ABSA model. The interface allows the user to type in names of colleges and select specific aspects, which will then provide a summary of the student’s feelings and the direction of those feelings for each area. This tool is highly easy to get sentiment analysis results and see important insights, which would be helpful for stakeholders to make decisions based on data. This study showed how important it is to understand student feelings in higher education. The DistilBERT-based ABSA model outperformed the traditional models and illustrated how effective transformer-based models can be in getting more detailed insights from student reviews. This research not only adds to the comprehension of student feedback but also opens avenues for the enhancement of decisions made by educational institutes.