Sentiment Analysis of University Social Media Using Support Vector Machine and Logistic Regression Methods

Abstract

Social media has become one of the most powerful platforms for information sharing. Colleges and universities now have official social media profiles to convey information about the campus and boost its branding and popularity. Instagram is a popular social networking website among college students. It is important for a university to comprehend its performance from the community's perspective, whether positive, negative, or indifferent toward the university. One solution is to examine the university's social media sentiment to establish the public's perception of the university. In this study, we will conduct a sentiment analysis on university social media based on public opinion or comments for each post on the university's Instagram to identify whether the comments are “Positive,” “Negative,” or “Neutral.” To classify posts on university Instagram, we use two methods: Support Vector Machine and Logistic Regression. The results suggest combining the Support Vector Machine approach with the TF-IDF feature yields the best F1-Score performance. In contrast, Logistic Regression with the FastText feature produces the worst performance of all models and feature extraction employed.