Sentimen Analisis Komentar Video Review Produk Tiktok Dengan Perbandingan Kinerja SVM, LSTM, Dan IndoBERT

The development of TikTok as a social media platform and digital commerce platform has led to increased user interaction through comments on product review content. These comments may contain various sentiments that can be used to identify users’ responses to a product. This study aims to analyze the sentiment of comments on product review videos on TikTok and compare the performance of the Support Vector Machine (SVM), Long Short-Term Memory (LSTM), and IndoBERT methods. The data consisted of 1,670 comments labeled using the Lexicon-Based method with the InSet lexicon and classified into three sentiment classes: negative, neutral, and positive. The evaluation was conducted using an 80:20 data split with accuracy, precision, recall, and F1-score as evaluation metrics. The results showed that positive sentiment was the most dominant, with 901 comments (53.95%), followed by neutral sentiment with 422 comments (25.27%) and negative sentiment with 347 comments (20.78%). Based on the evaluation results, IndoBERT achieved the highest accuracy of 78.44%, while SVM achieved an accuracy of 73.35% and LSTM achieved 77.84%. Therefore, IndoBERT demonstrated the best performance based on accuracy in this study.