De anonymizability of social network: through the lens of symmetry

Benjie Miao,Shuaiqi Wang,Luoyi Fu,Xiaojun Lin

DOI: https://doi.org/10.1145/3397166.3409127

2020-01-01

Abstract:As the popularity of social networks increases, the privacy of personal information in social networks becomes an issue of great concern. Concealing personal identity in social network is one of the most common methods to protect personal information, but it is insufficient for privacy protection since adversaries may use correlated side information across multiple networks to uncover the identity of anonymous user. Such re-identification process using auxiliary correlated information is called Social Network Deanonymization. The problem is initially proposed by Narayanan and Shimatikov [16]. In the past decades, a large number of works [16][19][14][11][3][13][9][18][24][4][20] have emerged focusing on the de-anonymization problem with different aspect. In this work, our focus is on an important branch of de-anonymization problem: seedless de-anonymization. In seedless de-anonymization, the attacker needs to re-identify the user identity in a published network using an auxiliary network with full identity information. The published network is completely anonymous, where no pre-identified nodes (ie the so-called seeds) are given. The correlation between two networks only lies in the similarity in their topology, since these two networks are supposed to be from the same underlying relationship network. The attacker aims to uncover the identities in this published anonymous network by matching the users in the published network to those in the auxiliary network.Various algorithms for seedless de-anonymization have been proposed [16][19][14][13][9][18][24][4]. Unfortunately, such algorithms may occasionally fail to perfectly de-anonymize the …