Collaborative filtering-based Madura Island tourism recommendation system using RecommenderNet

Abstract

Personalized tourism recommendations in Madura Island are limited, posing challenges for visitors in selecting destinations that match their preferences. This study develops a Collaborative Filtering-based recommendation system using a modified Cosine similarity approach combined with Convolutional Neural Networks (CNN) to improve personalized tourism suggestions for visitors to Madura Island. The investigation validated the system's accuracy in providing tailored recommendations by thoroughly exploring popular attractions and user preferences, highlighting its potential to enhance the overall tourism experience. User ratings for tourist attractions were collected via Google Forms. The collected data were processed, formatted, and modeled using the RecommenderNet architecture, with training and validation to optimize system performance. Data exploration revealed insights into key island attractions. The recommendation system yielded personalized suggestions, exemplified by a sample user's top-rated destinations. The system's performance was evaluated using RMSE, achieving a best score of 0.2579, demonstrating its accuracy in delivering personalized recommendations. The Collaborative Filtering-based recommendation system effectively improved user experiences by providing personalized tourism suggestions. Future work should focus on enhancing algorithmic approaches and expanding data integration to further refine and enrich tourism experiences in Madura Island.