Swarm intelligence for protecting sensitive identities in complex networks

The growing concern about data security in complex networks is a topic of immense interest. In this paper, we study how community deception can prevent malicious attackers from stealing confidential and private information from users. Community deception aims to tackle problems generated by community detection algorithms by rewiring the edges minimally. Finding the optimal edge set is essential in community deception, but this is a challenging problem coupled with high complexity. To address this issue, we propose the Swarm Community Protector (SCP), a swarm intelligence-based privacy-preserving model in terms of community structure. The SCP is designed to be adapted for graphs, and specifically edges, as opposed to the classical swarm algorithms, which are suitable for continuous and discrete domains only. We also develop an adaptive mechanism to ensure global and local searchability. A novel fitness function based on structure entropy is proposed to promote the obfuscation of community structure, thereby hiding the membership of sensitive entities in networks. Furthermore, we customize the solution space and integrate a pre-processing mechanism to reduce the search space. We have tested SCP on various real-world networks, and the experimental results indicate that SCP can provide excellent protection to the targeted community.