Enhancing support vector machine performance using particle swarm optimization for sentiment analysis

Abstract

Recently, social media has established itself as a leading platform in various sectors. Meanwhile, text extraction and sentiment analysis classification have attracted significant attention in research. Regrettably, traditional sentiment analysis often falls short of accurately capturing sentiment nuances. At the same time, machine learning has enabled more effective sentiment analysis, data mining, and classification, as well as the development of models that incorporate artificial intelligence. Therefore, the purpose of this study is to optimize sentiment analysis of public opinion in social media regarding Grand Prix motorcycle racing (MotoGP) and World Superbike (WSBK) events using machine learning and an optimized machine learning method. This study applies the support vector machine (SVM) machine learning method and enhances its performance through optimization by integrating it with the particle swarm optimization (PSO) algorithm. This study found that the SVM method achieved 80.15% accuracy, 75.63% recall, and 76.89% F1-score. In contrast, the SVM method combined with PSO achieves accuracies of 81.82%, 79.9%, and 79.62% for recall, precision, and F1-score, respectively, in classifying the sentiment of sporting events. The implications suggest that applying Hybrid SVM with PSO significantly enhances classification accuracy in sentiment analysis.