ASPECT-BASED SENTIMENT ANALYSIS (ABSA) TERHADAP ULASAN HOTEL BINTANG LIMA DI JAKARTA MENGGUNAKAN METODE FINE-TUNING INDOBERT

The hotel industry is a service sector that highly depends on customer experience and perception. In the digital era, online reviews have become a major source for prospective guests to assess the quality of hotel services. This study aims to conduct Aspect-Based Sentiment Analysis (ABSA) on customer reviews of five-star hotels in Jakarta using the Fine-Tuning IndoBERT method. A total of 2,499 reviews were collected from Google Reviews for five hotels: Fairmont Jakarta, Four Seasons Jakarta, The Langham Jakarta, Raffles Jakarta, and The Grove Suites by Grand Aston. Five aspects were analyzed—cleanliness, service, room comfort, food and beverages, and main facilities—which were manually labeled. The data underwent preprocessing and was divided into 80% training, 10% validation, and 10% testing sets. The model was trained for 20 epochs and evaluated using classification metrics. The results show an accuracy of 95%, a precision of 0.86, a recall of 0.80, and an F1-score of 0.82. Positive sentiment dominated all aspects, particularly service (84.8%) and food and beverages (83.2%). The Langham Jakarta received the most positive reviews, while the main facilities aspect was mostly associated with neutral sentiment. This approach is effective in understanding customer opinions and supporting improvements in hotel service quality.
Keywords: Aspect-Based Sentiment Analysis, IndoBERT, Five-Star Hotels, Google Reviews, Review Analysis