RRS: Review-Based Recommendation System Using Deep Learning for Vietnamese

Tourism, which includes sightseeing, relaxation, and discovery, is a fundamental aspect of human life. One of the most critical considerations for traveling is accommodation, mainly hotels. To improve the travel experience, we have presented a solution for building a recommendation model using Vietnamese and user data to suggest travelers choose the ideal hotel. Our data was collected from two well-known websites, Traveloka and Ivivu, and includes information about hotels in Vietnam and users’ feedback history, such as comments, ratings, and the names of users and hotels. We then preprocessed and labeled the inter-annotator agreement for various aspects, including service (0.89), infrastructure (0.84), sanitary (0.83), location (0.89), and attitude (0.83). Our recommendation model is built by using Collaborative Filtering and deep learning techniques. Furthermore, we suggest incorporating context vectors from tourists’ Vietnamese comments in the recommendation process. The context model is developed by using deep learning techniques to extract topics and sentiments from the words effectively. The results of our proposed model, as measured by the MSE, were 0.027, which is significantly better than a context-free model using the same parameters, which had an MSE of 0.061. Additionally, our Deep Learning, created using PhoBERT embedding, had an accuracy of 81