ASPECT-BASED SENTIMENT ANALYSIS (ABSA) TERHADAP MOBIL TESLA DALAM KOMENTAR YOUTUBE BERBAHASA INDONESIA MENGGUNAKAN METODE FINE-TUNING INDOBERT

This study aims to analyze public perception of Tesla cars using the Aspect-Based Sentiment Analysis (ABSA) approach by leveraging a fine-tuned IndoBERT model. The dataset consists of user comments from YouTube discussing Tesla vehicles, focusing on six main aspects: performance, design, technology, price, battery, and comfort. The analytical process includes text preprocessing, multi-aspect and multi-label data annotation, and training the IndoBERT model to classify sentiment into three categories: positive, neutral, and negative. The results show that the IndoBERT model is capable of effectively classifying aspect-based sentiment and capturing opinion context in user comments. Sentiment distribution indicates that most comments are neutral, but there are notable differences in the proportion of positive sentiment across aspects. The design aspect received the highest proportion of positive sentiment (7.4%), followed by technology (6.1%) and price (3.2%). These findings suggest that Tesla's visual design and aesthetics are the most appreciated factors by users, followed by technological sophistication and price perception. This research is expected to provide insights for automotive companies and policymakers to better understand consumer perception and formulate more targeted product development and communication strategies.
Keywords: Aspect-Based Sentiment Analysis, IndoBERT, YouTube Comments, Tesla, Sentiment Classification