Hiding in Temporal Networks

TL;DR

This paper explores how individuals can strategically rewire their connections over time in temporal networks to hide their importance from adversaries using temporal centrality measures, revealing computational challenges and privacy opportunities.

Contribution

It extends the study of social network privacy from static to temporal networks, analyzing the computational feasibility and structural factors influencing hiding strategies.

Findings

Optimal hiding strategies are usually computationally infeasible.

Temporal networks provide more manipulation strategies for privacy.

Temporal networks can be easier to hide in compared to static networks.

Abstract

Social network analysis tools can infer various attributes just by scrutinizing one's connections. Several researchers have studied the problem faced by an evader whose goal is to strategically rewire their social connections in order to mislead such tools, thereby concealing their private attributes. However, to date, this literature has only considered static networks, while neglecting the more general case of temporal networks, where the structure evolves over time. Driven by this observation, we study how the evader can conceal their importance from an adversary armed with temporal centrality measures. We consider computational and structural aspects of this problem: Is it computationally feasible to calculate optimal ways of hiding? If it is, what network characteristics facilitate hiding? This topic has been studied in static networks, but in this work, we add realism to the problem by considering temporal networks of edges changing in time. We find that it is usually computationally infeasible to find the optimal way of hiding. On the other hand, by manipulating one's contacts, one could add a surprising amount of privacy. Compared to static networks, temporal networks offer more strategies for this type of manipulation and are thus, to some extent, easier to hide in.