ASPECT-BASED SENTIMENT ANALYSIS ON TINDER APP REVIEWS USING THE SUPPORT VECTOR MACHINE METHOD

Abstract

The research was conducted to evaluate the performance of the Support Vector Machine (SVM) in classifying sentiment in Tinder app user reviews, focusing on four aspects: price, app features, account security, and non-aspect. Of the 1627 reviews analyzed, 1138 were used for training and 489 for testing. Performance evaluation was carried out using accuracy, precision, recall, and F1-score metrics. The results indicated that SVM performance varied depending on the aspect analyzed. The highest accuracy was achieved in the account security aspect, with a score of 0.944, while the lowest accuracy was found in the app features aspect, at 0.8998. The highest precision for negative and neutral sentiments was observed in the account security aspect, while precision for positive sentiment was the lowest. Conversely, the highest recall was found for neutral sentiment, particularly in the app features and account security aspects. However, recall for positive sentiment in the app features aspect was very low, indicating the model's difficulty in detecting positive reviews. Overall, SVM demonstrated good performance, especially in the account security aspect and neutral sentiment, but faced challenges in classifying positive sentiment in the app features aspect.