<![CDATA[The rapid growth of the hospitality industry and the increasing reliance on online reviews emphasize the need for advanced sentiment analysis tools to understand customer preferences effectively. This study explores the application of IndoBERT, a pre-trained language model tailored for the Indonesian language, in classifying sentiments from hotel guest reviews. Utilizing a dataset of 715 reviews, the study employed the Knowledge Discovery in Databases (KDD) framework for systematic data preprocessing, feature extraction, and machine learning analysis. IndoBERT demonstrated exceptional performance, achieving perfect precision, recall, and F1-scores of 1.00 for both positive (657 reviews) and negative (53 reviews) sentiment classes. The ROC curve analysis also yielded a mean AUC score of 0.86, validating the model's robustness and reliability. The results highlight IndoBERT's capability to accurately capture linguistic nuances and contextual meaning, offering actionable insights into factors influencing guest satisfaction, such as cleanliness, staff behavior, and service quality. This research contributes to advancing natural language processing applications in regional contexts and provides practical implications for enhancing service strategies in the hospitality sector. Future research should expand the model's application to other industries and explore multimodal approaches for a more comprehensive understanding of customer behavior.]]>
Copyrights © 2025
