Sentiment analysis on student well-being in Singapore

Abstract

The issue of well-being among students, particularly at the university level, has become increasingly important in recent years due to the growing awareness of mental health concerns among students. This project aims to analyze the sentiments and well-being of university students in Singapore by utilizing sentiment analysis tools and models on social media text. By examining general trends in sentiments, schools can identify areas that require attention and develop targeted solutions to address specific factors that contribute to poor well-being among students. Various sentiment analysis models were applied, including VADER, RoBERTa, and SenticNet, to classify polarity, while BART and SenticNet were used to classify well-being. The results indicated that RoBERTa had the highest accuracy rate of 77% for detecting polarity, while BART had the highest accuracy rate of 73% for detecting well-being.