Improvement of reading platforms assisted by the spring framework: A recommendation technique integrating the KGMRA...

With the widespread adoption of personalized recommendation systems, traditional methods continue to face significant challenges in areas such as recommendation accuracy, user experience, and content diversity. Existing approaches struggle to effectively integrate user behavior data with the semantic information of article content and often fall short in covering long-tail content to meet users' diverse needs. To address these limitations, this study introduces a novel recommendation system that combines the Knowledge Graph-based Multi-Relational Association (KGMRA) algorithm with the Bidirectional Encoder Representations from Transformers (BERT) model. The KGMRA algorithm leverages knowledge graphs to extract rich associative information and employs multi-relational networks to capture the semantic relationships between articles effectively. Concurrently, the BERT model uses deep learning to generate robust semantic representations of article content, enhancing the system's capacity to understand and predict user interests with greater precision. By integrating these advanced technologies, the proposed system achieves significant improvements in recommendation accuracy, personalization, and content diversity. Specifically, it excels in recommending long-tail content, thereby catering more effectively to users' interests in niche or less popular articles. Experimental results highlight the system's superior performance compared to existing baseline methods. Key metrics improved substantially, with recommendation accuracy increasing from 72 % to 84 %, coverage rising from 71 % to 89 %, and the click-through rate growing from 79 % to 94 %. Additionally, the system's efficiency improved by 20 %, resulting in faster response times and an enhanced user experience. This study provides a practical and effective solution for improving recommendation systems on reading platforms built with the Spring framework. It not only offers significant application value but also contributes new perspectives for the optimization and innovation of future recommendation systems.