Obfuscation of Node Centrality in Networks

Centrality measures determined by the network’s topology structure (e.g., degree centrality, eigenvector centrality, and betweenness centrality) are widely utilized in various fields, such as network analysis and data mining. While extensive research has proposed several centrality indices to measure how important a node is compared to others as well as fast centrality computation algorithms, there is a demand to obfuscate node centrality from a security and privacy perspective. Obfuscating node centrality offers potential applications, such as preventing attacks on important nodes from malicious actors. Therefore, this paper aims to develop an algorithm for obfuscating the centrality of arbitrary sets of nodes in a graph through the addition and deletion of links. Specifically, we introduce COBF (Centrality OBFuscation), which applies the concept of ProHiCo (Probabilistic algorithm to Hide Communities), a community obfuscation method proposed, to obfuscate node centrality. Experiments are conducted to analyze how effectively the centrality of specified nodes is obfuscated. The results demonstrate that COBF can effectively alter node rankings by approximately five positions in a network with a total of $\mathbf{1 0 0}$ nodes, highlighting its potential for enhancing privacy in network analysis.