Abstract: DP-GSPerturb is a distributed personalized social privacy protection method based on graph structure perturbation; it is proposed to solve sensitive link privacy issues in personalized social networks. The method is a node-centric method that searches reachable nodes of sensitive source nodes, transfers reachable information to sensitive source nodes, and randomly perturbs links of sensitive source nodes through between-node messaging, node value updating to achieve the personalized privacy protection of sensitive link in the distributed environment. The experimental results show that DP-GSPerturb improves not only the processing speed of large-scale graph data but also the availability of data published.