Personalized tourist attraction recommendation using graph neural networks, collaborative filtering, and optical data modeling

The rapid development of online travel platforms and the surge in user-generated content have created a wealth of data, but they also pose challenges for travelers in selecting destinations that align with their preferences. To address these challenges, this study proposes a hybrid recommendation framework that integrates Graph Neural Networks (GNNs) with Collaborative Filtering (CF) and incorporates optical data modeling. Optical data, such as visual information from images and light-based interaction analysis, is utilized to enhance the feature representation of tourist attractions, providing an additional layer of contextual information. The framework leverages GNNs to model high-order dependencies within user-item interaction networks, enabling the extraction of implicit relationships, while CF incorporates historical behavioral patterns to enhance recommendation accuracy. A bipartite graph representation is constructed to encode user-item interactions, and a multi-layer GNN architecture is employed to propagate and aggregate feature information. The CF module refines the recommendations by modeling user and item biases. This integration of optical techniques addresses data sparsity and improves the robustness of the recommendations, making it a scalable and adaptive solution for intelligent tourism platforms.