Secure and personalized sports media recommendation via federated knowledge graph embedding in edge computing

With the rapid growth of digital sports media services, users increasingly demand personalized content experiences that respect privacy and provide secure recommendations. Traditional centralized recommendation methods face critical challenges, such as user data privacy risks, bandwidth limitations, and latency issues, particularly when processing extensive multimedia data. To overcome these problems, this paper proposes a novel framework integrating federated knowledge graph embedding (FKGE) with edge computing for secure and personalized sports media recommendations. Specifically, FKGE allows collaborative learning of sports media preferences by embedding user and content information into a distributed semantic space without exposing raw data. Edge computing nodes are employed to locally process and store knowledge graphs, significantly reducing response latency and alleviating network load. Experimental evaluations demonstrate that the proposed framework not only achieves higher recommendation accuracy compared to existing methods but also effectively protects user privacy and enhances user experience in personalized sports media content delivery.