Personalized Book Recommendation Based on Relational Graph Convolutional Network

As libraries manage increasingly vast collections, the task of identifying suitable books for readers has become increasingly time-consuming and complex. Traditional recommendation systems face challenges in effectively handling networks charac-terized by intricate interactions and diverse entity relationships, often struggling with scalability. To address these limitations, this study explores the use of Relational Graph Convolutional Networks (RGCN) to improve book recommendation systems. A knowledge graph is constructed to capture the complex relation-ships between entities such as readers, books, and their various attributes. Through the RGCN architecture, the model learns relation-specific embeddings, enabling accurate link prediction for book recommendations. Experimental results demonstrate that this approach not only surpasses traditional recommendation systems in terms of accuracy and scalability but also offers valuable insights into reader preferences and book relevance, thereby enhancing the recommendation process in large-scale library systems.