Abstract:Social recommendation refers to recommendation technology taking social relations as additional input to improve merchandise sales and user satisfaction. It has been widely used in various social networking services. Social recommendation requires the collection of each user's social relations to build a social graph, posing the risk of user privacy intrusion caused by untrusted servers. Local differential privacy is a widely adopted approach for providing privacy protection while allowing acceptable utility of the protected data for analytics. Growing research interest has been applying local differential privacy protection to social graph analysis. However, local differential privacy is suitable for analyzing coarse-grained statistics from perturbed user data, and it is difficult to obtain the fine-grained user relations needed for social recommendation. This paper proposes a novel locally differentially private social recommendation method. It first perturbs the degrees of users in a given interval and discovers coarse-grained communities from them, then generates a fine-grained social graph according to the features of intra-community and inter-community relations, thus ensuring the accuracy of the recommendation results. Experimental results show that the social recommendation accuracy of our proposed method improves by about 20% compared to existing methods in a community-structured ego-network dataset, and by about 10% in two non-community-structured page-page network datasets. The experimental results show that compared with the state-of-the-art methods such as LDPGen, LFGDPR, and AsgLDP, the social recommendation accuracy of our proposed method is improved by about 20% in one community-structured ego-network dataset, and 10% in two non-community-structured page-page network datasets.

Graph Neural Networks for Privacy-Preserving Recommendation with Secure Hardware

Decentralized Graph Neural Network for Privacy-Preserving Recommendation

Subgraph-level federated graph neural network for privacy-preserving recommendation with meta-learning

A Privacy-Preserving Subgraph-Level Federated Graph Neural Network via Differential Privacy

Devil in Disguise: Breaching Graph Neural Networks Privacy through Infiltration

FedGNN: Federated Graph Neural Network for Privacy-Preserving Recommendation

A GCN-based Model for Recommendation Using Local Differential Privacy Method

Multi-Party Secure Computation with Intel SGX for Graph Neural Networks

Privacy-Enhanced Graph Neural Network for Decentralized Local Graphs

Differentially Private Graph Neural Networks for Link Prediction

Privacy-Preserving Recommendation Based on a Shuffled Federated Graph Neural Network

Privacy-Preserving Decentralized Inference with Graph Neural Networks in Wireless Networks

Towards Differential Privacy in Sequential Recommendation: A Noisy Graph Neural Network Approach

SecGNN: Privacy-Preserving Graph Neural Network Training and Inference as a Cloud Service

Community-based Social Recommendation under Local Differential Privacy Protection

P4GCN: Vertical Federated Social Recommendation with Privacy-Preserving Two-Party Graph Convolution Networks

Achieving Adaptive Privacy-Preserving Graph Neural Networks Training in Cloud Environment

Private and Utility Enhanced Recommendations with Local Differential Privacy and Gaussian Mixture Model

Efficient Privacy Preserving Graph Neural Network for Node Classification

Individual Fairness for Local Private Graph Neural Network