Federated knowledge graph-based hybrid personalized curriculum recommendation for privacy-preserving physical education teaching systems

With the growing emphasis on personalized instruction in physical education (PE), there is an increasing need to dynamically tailor learning paths according to individual abilities and progression. However, achieving such personalization often relies on collecting and processing sensitive learner data, raising serious privacy concerns. Furthermore, PE curricula typically involve complex prerequisite structures and evolving learner profiles, which are difficult to capture with conventional recommendation techniques. In this work, we present a federated knowledge graph-based hybrid framework for adaptive and personalized curriculum recommendation in physical education teaching systems. Our approach enables multiple institutions to collaboratively build and refine knowledge representations without sharing raw learner data. By integrating domain-specific knowledge graphs with a federated training protocol, the system effectively models relationships among learners, skills, and instructional content while preserving privacy. This framework lays the groundwork for developing PE teaching systems that are both adaptive to individual needs and compliant with privacy requirements.