A Privacy-Preserving Distributed Contextual Federated Online Learning Framework with Big Data Support in Social Recommender Systems

Nowadays the booming demand of big data analytics and the constraints of computational ability and network bandwidth have made it difficult for a stand-alone agent/service provider to provide suitable information for every user from the large volume online data within the time limited. To handle this challenge, a recommender system (RS) can call in a group of agents to collaborate to learn users' preference and taste, which is known as a distributed recommender system (DRS). In this paper, we propose a privacy-preserving DRS, where each service provider (SP) is modeled as a distributed context-aware online learner. SPs collaborate to make personalized recommendations by learning users' preferences based on the user context and users' previous behaviors. To support big data analytics, we establish an item-cluster tree from top to the bottom to handle increasing datasets. We further consider the structure of social network and propose an adaptive algorithm to reduce performance loss. Theoretical analysis show that our proposal can achieve sublinear regret and differential privacy of both SPs and users. Numerical results validate that our new framework can support increasing big datasets while striking a balance between privacy-preserving level (PPL) and prediction accuracy.