A Resource Recommendation and Sharing Method for Virtual Teaching and Research Offices Based on Personalized Federated Learning

The virtual teaching and research office, as a new organizational form in education, relying on information technology, breaks through time and space, and promotes interdisciplinary and interregional collaboration and resource sharing. It has emerged as a significant force in higher education reform. However, in the process of resource sharing and data exchange in its practical application, the virtual teaching and research office still has problems such as privacy leakage and data island effect, which restrict its broader application. As a result, the premise of the effective sharing of educational resources and the protection of data privacy and security have become the major problems in the development of teaching and research offices of virtual universities. In order to overcome this challenge, the present study proposes a solution through a personalized federated learning (pFL-VAR)-based resource-sharing method for virtual teaching and research office. It introduces federated learning technology to protect the privacy of teaching data and resources and improve the accuracy of user personalized recommendations. Concretely, pFL-VAR supplies a personalized model for every unit of the virtual teaching and research office, and realizes the knowledge share between personalized models through prototype exchange, thus improving the accuracy and izability of personalized recommendation models. Experiments on MOOC data set from University of Electronic Science and Technology of China show that pFL-VAR has a strong personalized recommendation ability, which provides an important basis for sharing resources in virtual teaching and research offices.