Hiding in Multilayer Networks

Multilayer networks allow for modeling complex relationships, where individuals are embedded in multiple social networks at the same time. Given the ubiquity of such relationships, these networks are increasingly gaining attention in the literature. We present the first analysis of the robustness of centrality measures against strategic manipulation in multilayer networks. Specifically, we consider an evader who wishes to score low in a centrality-based ranking - to avoid being highlighted as a key figure in the network - while ensuring that she is connected to a certain group of people. In this context, we prove that determining the optimal way to ``hide'' is NP-complete in most cases. Given this, we empirically evaluate a number of heuristic algorithms that the evader may use, and show that the most challenging centrality measures to hide from are those that require global information about the entire network topology.