Bipartite Graph-Based Recommendation Systems: Introducing the Enhanced Graph Convolutional Neural Network Method for Indonesian Tourism

Indonesia's tourism sector, a cornerstone of its economy, has seen significant growth with both domestic and international tourists, drawn by its diverse landscapes and cultural sites. In 2022, the country welcomed around 10 million international visitors, indicating the industry's potential. In this context, employing artificial intelligence for personalized travel recommendations is crucial. This study proposes an innovative approach, converting tourism data into a bipartite format focused on the user-place relationship. We introduce the Enhanced Graph Convolutional Neural Network (Enhanced GCN), tailored for the tourism sector's complexities. Compared to the traditional Graph Convolutional Network (GCN), the Enhanced GCN shows superior predictive accuracy, reducing the Mean Squared Error (MSE) to 2.38. This advance represents a significant step in providing more accurate, customized travel suggestions. However, the current MSE suggests room for further improvement. Future research will aim to refine the Enhanced GCN, enhancing its ability to meet and predict tourists' needs in Indonesia's dynamic tourism environment. Ultimately, the goal is to use artificial intelligence to transform how visitors experience Indonesia's cultural and natural offerings, thereby bolstering the tourism industry's growth..