Digital Traces in Tourism: Leveraging NLP to Evaluate Tourist Experiences Across Southeast Asian Destinations

The tourism industry significantly influences the global economy. Digital reviews have become crucial in guiding tourists to their preferred destinations as technology advances. However, the inherently textual nature of reviews presents challenges for traditional processing methods, necessitating specialized approaches to handle the complexities. This research leverages Natural Language Processing (NLP) techniques to examine visitor feedback for well-known tourist destinations in Indonesia and Thailand. We employ a transformer-based deep learning model, BERT, to perform sentiment analysis and multiclass classification on the collected reviews. The sentiment analysis results reveal high levels of visitor satisfaction for both countries, with 77.18% positive sentiment for Indonesian destinations and 78.52% for Thai tourist spots. We applied multiclass text classification techniques to identify the primary dimensions of perceived tourist experiences. Entertainment emerged as the main reason tourists visit destinations in Indonesia and Thailand. Each country offers distinct favored entertainment experiences, such as nature-based entertainment in Indonesia and nightlife and festive entertainment in Thailand. This study provides valuable insights for government authorities and tourism stakeholders on the perception of tourist experiences, supporting and enhancing the attractiveness of destinations in both countries.