<![CDATA[Guest reviews on Online Travel Agents have become an important source for assessing hotel service quality, yet conventional sentiment analysis only produces an overall score and fails to capture opinions on specific service aspects, making it difficult to determine improvement priorities. This study aims to build an Aspect-Based Sentiment Analysis (ABSA) framework that turns reviews into actionable service-improvement recommendations. Reviews from several platforms were collected, labeled, and classified using a fine-tuned IndoBERT model to identify seven service aspects along with their sentiment polarity, after which the results were summarized and presented through an interactive monitoring dashboard. The proposed model achieved 92.83% accuracy, 92.89% weighted F1, and 74.55% macro F1. These results show that the proposed framework delivers more interpretable and targeted insights than conventional sentiment analysis, thereby supporting data-driven decision-making for service improvement in the hospitality industry. Keywords: Aspect-Based Sentiment Analysis; IndoBERT; Hotel Reviews; Monitoring Dashboard; Online Travel Agent Abstrak Ulasan tamu di Online Travel Agent menjadi sumber penting untuk menilai kualitas layanan hotel, namun analisis sentimen konvensional hanya menghasilkan skor keseluruhan dan gagal menangkap opini pada aspek layanan tertentu, sehingga sulit menentukan prioritas perbaikan. Penelitian ini bertujuan membangun kerangka kerja Analisis Sentimen Berbasis Aspek (ABSA) yang mengubah ulasan menjadi rekomendasi perbaikan layanan yang dapat ditindaklanjuti. Ulasan dari beberapa platform dikumpulkan, dilabeli, lalu diklasifikasikan menggunakan model IndoBERT yang disempurnakan untuk mengenali tujuh aspek layanan beserta polaritas sentimennya, kemudian hasilnya diringkas dan disajikan melalui dashboard pemantauan interaktif. Model yang diusulkan mencapai akurasi 92,83%, F1 tertimbang 92,89%, dan F1 makro 74,55%. Hasil ini menunjukkan bahwa kerangka kerja yang diusulkan memberikan wawasan yang lebih interpretatif dan tepat sasaran dibanding analisis sentimen konvensional, sehingga mendukung pengambilan keputusan perbaikan layanan berbasis data di industri perhotelan.]]>
