Evaluation of Sentiment Analysis Methods for Social Media Applications: A Comparison of Support Vector Machines and Naïve Bayes

Abstract

This study compares the effectiveness of the Support Vector Machine (SVM) and Naïve Bayes methods in sentiment analysis of TikTok application reviews in Indonesia. The primary objective is determining which method better classifies positive and negative sentiments. The dataset consists of TikTok reviews collected from Indonesian users. SVM and Naïve Bayes methods are evaluated using accuracy, precision, recall, F1-score, and Area Under the Curve (AUC). The results show that SVM outperforms Naïve Bayes in detecting positive sentiments, with higher precision, significant recall, and a more robust F1-score. SVM’s AUC further highlights its ability to differentiate between positive and negative reviews. While Naïve Bayes offers some advantages in specific cases, SVM is recommended for applications requiring more precise sentiment detection on social media platforms like TikTok. The practical implications of this research are considerable. First, the findings can help developers and data analysts improve automated sentiment analysis tools, leading to better accuracy in classifying user feedback. Second, content moderation systems can leverage SVM to identify and mitigate harmful content, enhancing users' overall safety and experience more effectively. Third, businesses can utilize these insights to optimize their marketing strategies, tailoring campaigns based on real-time sentiment analysis. These applications will improve user engagement, reputation management, and customer satisfaction. Future research should explore additional machine learning techniques and further refine sentiment analysis models for enhanced performance.

Topics & Concepts