SVM Optimization with Kernel Function for Sentiment Analysis on Social Media twitter (X) in AFC U23 Asian Cup Case Study

Abstract

This study aims to optimize the performance of the Support Vector Machine (SVM) in sentiment analysis on social media by using various kernel functions, namely linear, polynomial, and Radial Basis Function (RBF). The case study taken was a conversation related to the AFC Asian Cup U-23 taken from social media platforms. The data used in this study included three classes of sentiment: positive, neutral, and negative. The experimental results show that the linear kernel achieves the highest accuracy of 93.55% with an F1-score of 0.9296. The RBF kernel shows almost comparable performance with an accuracy of 90.05% and an F1-score of 0.8820. In contrast, the polynomial kernel showed lower performance with an accuracy of 80.65% and an F1-score of 0.7346. The results of the analysis using the confusion matrix show that linear kernels and RBF are more effective in classifying neutral and positive sentiment than polynomial kernels. This study confirms that the right selection of kernels in SVM greatly affects the accuracy and effectiveness of sentiment analysis. Linear kernels and RBFs have proven superior in handling complex sentiment analysis datasets, such as those related to the AFC U-23 Asian Cup. These findings can be used as a basis for further development in sentiment analysis applications across various domains.