Burning some myths on privacy properties of social networks against active attacks

2025-04-17

AI Summary

This paper challenges the prevailing assumption that social networks are inherently privacy-vulnerable, focusing on structured privacy modeling and enhancement under active adversarial attacks. Method: We propose the $(k,ell)$-anonymity metric and the $k$-degree-measure-resistant graph ($k$-MDRG) theoretical framework; design graph-embedding-based privacy-enhancing transformation operations; and establish formal properties and constructive methods for $k$-MDRGs with $k geq 1$. Contribution/Results: We prove—first time—that real-world social networks are almost universally *not* $1$-MDRGs, thereby refuting the “inherent extreme vulnerability” hypothesis. Experiments confirm that $1$-MDRGs are negligible in random graphs and absent across all tested real social networks. Our framework provides a formalizable, augmentable paradigm for social graph privacy, enabling systematic resilience against active adversaries through principled structural perturbation and quantifiable anonymity guarantees.

Technology Category

• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Machine Learning: Privacy
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

This work focuses on showing some arguments addressed to dismantle the extended idea about that social networks completely lacks of privacy properties. We consider the so-called active attacks to the privacy of social networks and the counterpart $(k,ell)$-anonymity measure, which is used to quantify the privacy satisfied by a social network against active attacks. To this end, we make use of the graph theoretical concept of $k$-metric antidimensional graphs for which the case $k=1$ represents those graphs achieving the worst scenario in privacy whilst considering the $(k,ell)$-anonymity measure. As a product of our investigation, we present a large number of computational results stating that social networks might not be as insecure as one often thinks. In particular, we develop a large number of experiments on random graphs which show that the number of $1$-metric antidimensional graphs is indeed ridiculously small with respect to the total number of graphs that can be considered. Moreover, we search on several real networks in order to check if they are $1$-metric antidimensional, and obtain that none of them are such. Along the way, we show some theoretical studies on the mathematical properties of the $k$-metric antidimensional graphs for any suitable $kge 1$. In addition, we also describe some operations on graphs that are $1$-metric antidimensional so that they get embedded into another larger graphs that are not such, in order to obscure their privacy properties against active attacks.

Problem

• Debunking myths about social networks lacking privacy properties
• Analyzing active attacks and (k,ℓ)-anonymity in social networks
• Investigating k-metric antidimensional graphs for privacy quantification

Innovation

• Uses (k,ℓ)-anonymity for privacy quantification
• Analyzes k-metric antidimensional graphs theoretically
• Employs graph operations to obscure privacy properties