IMPLEMENTATION OF SVM-PSO IN SENTIMENT ANALYSIS OF GOOGLE PLACE REVIEW USERS AT CAFE HEADQUARTERS

Sentiment analysis plays an important role in understanding customer perceptions of businesses, allowing companies to respond more effectively to customer needs and satisfaction. This study aims to evaluate the performance of a Support Vector Machine (SVM) model optimized with Particle Swarm Optimization (PSO) in classifying the sentiment of user reviews on Markas Cafe. The dataset consists of 1,533 user reviews categorized into three sentiment classes: positive, neutral, and negative. The optimization process using PSO is used to find the optimal SVM parameters. The results showed that the SVM-PSO model achieved an accuracy of 87.7% and an Area Under Curve (AUC) of 0.85, with the best performance on positive sentiment (94.7% precision and 92.8% recall). Although the model showed good ability in detecting positive sentiments, the results for neutral and negative sentiments indicated the need for further improvement. This study confirms the effectiveness of SVM-PSO in sentiment analysis and suggests this approach can be utilized by businesses to improve marketing and customer service strategies based on user feedback.