Degrading the accuracy of interlayer link prediction: A method based on the analysis of node importance

Abstract

Interlayer link prediction aims to identify the same entities in multiplex networks, providing significant value for applications like social network analysis and personalized recommendations. However, malicious individuals can misuse it to aggregate users’ private information. To address this, social network platforms must explore preventive measures, such as useful information deletion or noise information addition, to attack successful interlayer link prediction. This study introduces a novel node importance-based network structure perturbation method to degrade the accuracy of interlayer link prediction algorithms hence protecting user privacy. It utilizes the “guessing the friends” game analogy to identify influential nodes in social networks based on degree centrality. Perturbation involves randomly removing connections from nodes with low, medium, and high degree centrality. Multiple interlayer link prediction algorithms are evaluated under these perturbation attacks to assess their impact. Experiments on real-world datasets validate the proposed method’s effectiveness and reveal that perturbing the nodes with low degree centrality significantly affects interlayer link prediction accuracy. This research contributes valuable insights into safeguarding user privacy in interlayer link prediction, offering a robust approach anchored in node importance considerations. It promises to advance interlayer link prediction techniques and protect social network user privacy effectively.