Enhancing user insights: a visualization dashboard for aspect-based sentiment analysis of Malaysia's island travel agencies

Island travel agencies are growing in popularity in Malaysia, leading to increased competition. Traveler satisfaction is crucial for business success, and platforms like Google Maps and Facebook provide valuable reviews for evaluation. However, these reviews are often unstructured and difficult to analyze. This project aims to develop a web application using Aspect-Based Sentiment Analysis (ABSA) to classify and visualize reviews for island travel agencies in Malaysia. The system employs Naïve Bayes (NB) for sentiment classification and Latent Dirichlet Allocation (LDA) for aspect detection, focusing on four categories: price, guide, experience, and service. Results are displayed through pie charts, word clouds, bar charts, and line charts. The system was tested for accuracy, functionality, and usability. LDA achieved a coherence score of 0.428, while NB achieved 93.40% accuracy for English and 77.32% for Malay. Usability testing was conducted using the Usability Metric for User Experience (UMUX), yielding an overall satisfaction score of 92.50%, with high ratings for effectiveness, satisfaction, and efficiency, indicating exceptional usability. Future improvements include extending data scraping, and incorporating data from platforms like Trivago and Agoda. The system is highly effective, user-friendly, and provides valuable insights for travelers and agencies.