RecGuard: A Blockchain-based Privacy Preservation System for Safeguarding Personal Data of Online Social Network Users with Enhanced Efficiency and Security

Online social networks (OSNs) produce large volumes of user-generated data, enabling personalized services but also exposing users to significant privacy risks, a lack of transparency and frequent security breaches. Existing blockchain- and machine learning–based privacy-preservation methods struggle with high computational costs, limited scalability and weak malicious-node detection. To address these gaps, this work proposes a Blockchain-Driven Privacy Preservation Scheme with Progressive Graph Convolutional Networks (BPPS-SPD-PGCN) for secure and efficient protection of personal data in OSNs. The framework integrates Adaptive Two-Stage Unscented Kalman Filtering for data preprocessing, PGCN for malicious-node detection, ARPO for optimizing PGCN weights and Fair Proof-of-Reputation blockchain for secure access control. Two smart contracts (RG-SH and RG-ST) further enhance data confidentiality and storage integrity. Using the Epinions dataset, the proposed technique was evaluated through Accuracy, Precision, Recall, F1-score and Computational Time. The system achieved 99.04% Accuracy, 92.34% Precision, 99.14% Recall and 99.93% F1-score, outperforming PPB-OSN-GCN, HCS-PSC-SVM and BDI-ISPP-CNN. Overall, BPPS-SPD-PGCN provides a more robust, precise and secure privacy-preservation solution for OSNs, offering significant improvements over existing approaches.