Lexicon-Based, Naïve Bayes, C4.5 for Analyzing Visitor Data Reviews as Recommendations for Priority Development of Labuan Bajo Tourism

Labuan Bajo is one of Indonesia’s national priority destinations, known for its natural beauty and unique ecosystem. However, several issues raised by tourists—such as complaints regarding prices, infrastructure, and environmental cleanliness—may affect its image and tourism sustainability. This study aims to analyze tourist perceptions of Labuan Bajo based on 7,000 reviews from TripAdvisor and Google Maps obtained through web crawling. Sentiment labeling was conducted using a Lexicon-Based approach, while classification was performed using the Naïve Bayes and C4.5 algorithms, both with and without the Synthetic Minority Oversampling Technique (SMOTE). The results showed 4,374 positive, 2,769 negative, and 1,804 neutral reviews. Based on the CRISP-DM method, Naïve Bayes achieved the highest accuracy of 88%, compared to 78% for C4.5. Dominant positive terms such as beautiful, stunning, and sustainable highlight Labuan Bajo’s natural strengths, while negative terms like price, toilet, and trash indicate areas requiring improvement. The findings provide strategic recommendations to enhance tourism management and service quality toward sustainable tourism development.