Social Network De-anonymization Research Articles

On Social Network De-Anonymization With Communities: A Maximum A Posteriori Perspective

A crucial privacy-driven issue nowadays is re-identifying anonymized social networks by mapping them to correlated cross-domain auxiliary networks. Prior works are typically based on modeling social networks as random graphs representing users and their relations, and subsequently quantify the quality of mappings through varied cost functions. However, many cost functions are empirically proposed without sufficient theoretical support. For some other works probing the theoretical bound, it remains unknown how to algorithmically meet the demand of such quantifications, i.e., to minimize the cost functions. Besides, only few prior works have discussed the de-anonymization of social networks with communities. We address those concerns in a social network modeling parameterized by community structures that can be leveraged as side information for de-anonymization. Based on the Maximum A Posteriori (MAP) estimation, our first contribution is a series of MAP-based cost functions, which, when minimized, enjoy superiority to previous ones in finding the correct mapping with the highest probability. The feasibility of the cost functions is then for the first time algorithmically characterized. We prove the general multiplicative inapproximability and thus propose two heuristics, which, respectively, enjoy an <inline-formula><tex-math notation="LaTeX">$\epsilon$</tex-math></inline-formula> -additive approximation and a conditional optimality in carrying out successful user re-identification. Our theoretical findings are also empirically validated under classical synthetic and real-wrold social networks. Both theoretical and empirical observations manifest the importance of community in enhancing privacy inferencing.

Partial Recovery in the Graph Alignment Problem

Partially Recovering a Graph Alignment in the Correlated Erdös–Renyi Model Given two graphs, how can we partially recover a one-to-one mapping between nodes that maximizes edge overlap? This problem, known as the graph alignment problem, appears in settings such as social network deanonymization and cellular biology. In “Partial Recovery in the Graph Alignment Problem,” G. Hall and L. Massoulié consider a stylized mathematical model of problems of this type: they assume that the input graphs are generated via a probabilistic model, namely, the correlated Erdös–Renyi model with parameters (n, q, s). They provide both necessary and sufficient conditions on (n, q, s) under which partial recovery can be achieved. In particular, they show that partial recovery can be achieved in the nqs = Ɵ(1) regime under certain additional assumptions.

Fundamental Privacy Limits in Bipartite Networks Under Active Attacks

This work considers active deanonymization of bipartite networks. The scenario arises naturally in evaluating privacy in various applications such as social networks, mobility networks, and medical databases. For instance, in active deanonymization of social networks, an anonymous victim is targeted by an attacker (e.g. the victim visits the attacker’s website), and the attacker queries her group memberships (e.g. by querying the browser history) to deanonymize her. In this work, the fundamental limits of privacy, in terms of the minimum number of queries necessary for deanonymization, is investigated. A stochastic model is considered, where 1) the bipartite network of group memberships is generated randomly; 2) the attacker has partial prior knowledge of the group memberships; and 3) it receives noisy responses to its real-time queries. The bipartite network is generated based on linear and sublinear preferential attachment, and the stochastic block model. The victim’s identity is chosen randomly based on a distribution modeling the users’ risk of being the victim (e.g. probability of visiting the website). An attack algorithm is proposed which builds upon techniques from communication with feedback, and its performance, in terms of expected number of queries, is analyzed. Simulation results are provided to verify the theoretical derivations.

Collective De-Anonymization of Social Networks With Optional Seeds

As Internet users interacting with their different friends in different social networks, the de-anonymization problem has been raising improving concern. Since the assailants may de-anonymize a social network by matching it with a correlated sanitized network and identifying anonymized user identities, multifarious arts study on the theoretical conditions or practical algorithms for correctly de-anonymizing a social network. Except for the structural information of these social networks, there has also been bounteous works taking advantage of some pre-identified seed nodes for reference in the anonymized network. In this paper, we systematically probe the theoretical conditions and algorithmic approaches for correctly matching two different-sized social networks by leveraging the multi-hop neighborhood relationships. A limited number of seeds are also taken into consideration as auxiliary information. To this end, we introduce the de-anonymization problem with the aid of the collectiveness and the collective adjacency disagreements, which are the collection of disagreements of different multi-hop adjacency matrices. We theoretically demonstrate that minimizing the collective adjacency disagreements can help match two social networks even in a very sparse circumstance, as it significantly enlarges the difference between the mismatched node pairs and the correctly matched pairs. Besides, the seeds is proved to bring positive influence in improving the de-anonymization accuracy. Algorithmically, we relax the domain of the matching function to continuum and adopt the conditional gradient descending method on the collective-form objective, to efficiently minimize the collective adjacency disagreements of two networks. We conduct tremendous experiments on different networks with or without seeds, the results of which return desirable de-anonymization accuracies and reveal the advantages of the collectiveness: the collectiveness manifests rich structural information, thereby most nodes can be correctly matched with their correspondences even in some sparse networks, where merely utilizing the 1-hop adjacency relationships might fail to work.

Social Network De-Anonymization and Privacy Inference with Knowledge Graph Model

Social network data is widely shared, transferred and published for research purposes and business interests, but it has raised much concern on users’ privacy. Even though users’ identity information is always removed, attackers can still de-anonymize users with the help of auxiliary information. To protect against de-anonymization attack, various privacy protection techniques for social networks have been proposed. However, most existing approaches assume specific and restrict network structure as background knowledge and ignore semantic level prior belief of attackers, which are not always realistic in practice and do not apply to arbitrary privacy scenarios. Moreover, the privacy inference attack in the presence of semantic background knowledge is barely investigated. To address these shortcomings, in this work, we introduce knowledge graphs to explicitly express arbitrary prior belief of the attacker for any individual user. The processes of de-anonymization and privacy inference are accordingly formulated based on knowledge graphs. Our experiment on data of real social networks shows that knowledge graphs can power de-anonymization and inference attacks, and thus increase the risk of privacy disclosure. This suggests the validity of knowledge graphs as a general effective model of attackers’ background knowledge for social network attack and privacy preservation.

Social Network De-anonymization

Previous works on social network de-anonymization focus on designing accurate and efficient de-anonymization methods. We attempt to investigate the intrinsic relationship between the attacker’s knowledge and the expected de-anonymization gain. A common intuition is that more knowledge results in more successful de-anonymization. However, our analysis shows this is not necessarily true if the attacker uses the full background knowledge for de-anonymization. Our findings leave intriguing implications for the attacker to make better use of the background knowledge for de-anonymization and for the data owners to better measure the privacy risk when releasing their data to third parties.

Fast De-anonymization of Social Networks with Structural Information

Ever since the social networks became the focus of a great number of researches, the privacy risks of published network data have also raised considerable concerns. To evaluate users’ privacy risks, researchers have developed methods to de-anonymize the networks and identify the same person in the different networks. However, the existing solutions either require high-quality seed mappings for cold start, or exhibit low accuracy without fully exploiting the structural information, and entail high computation expense. In this paper, we propose a fast and effective seedless network de-anonymization approach simply relying on structural information, named RoleMatch. RoleMatch equips with a new pairwise node similarity measure and an efficient node matching algorithm. Through testing RoleMatch with both real and synthesized social networks, which are anonymized by several popular anonymization algorithms, we demonstrate that the RoleMatch receives superior performance compared with existing de-anonymization algorithms.

Understanding structure-based social network de-anonymization techniques via empirical analysis

The rapid development of wellness smart devices and apps, such as Fitbit Coach and FitnessGenes, has triggered a wave of interaction on social networks. People communicate with and follow each other based on their wellness activities. Though such IoT devices and data provide a good motivation, they also expose users to threats due to the privacy leakage of social networks. Anonymization techniques are widely adopted to protect users’ privacy during social data publishing and sharing. However, de-anonymization techniques are actively studied to identify weaknesses in current social network data-publishing mechanisms. In this paper, we conduct a comprehensive analysis on the typical structure-based social network de-anonymization algorithms. We aim to understand the de-anonymization approaches and disclose the impacts on their application performance caused by different factors, e.g., topology properties and anonymization methods adopted to sanitize original data. We design the analysis framework and define three experiment environments to evaluate a few factors’ impacts on the target algorithms. Based on our analysis architecture, we simulate three typical de-anonymization algorithms and evaluate their performance under different pre-configured environments.

Hiding information against structural re-identification

Connections between users of social networking services pose a significant privacy threat. Recently, several social network de-anonymization attacks have been proposed that can efficiently re-identify users at large scale, solely considering the graph structure. In this paper, we consider these privacy threats and analyze de-anonymization attacks at the model level against a user-controlled privacy-enhancing technique called identity separation. The latter allows creating seemingly unrelated identities in parallel, even without the consent of the service provider or other users. It has been shown that identity separation can be used efficiently against re-identification attacks if user cooperate with each other. However, while participation would be crucial, this cannot be granted in a real-life scenario. Therefore, we introduce the y-identity model, in which the user creates multiple separated identities and assigns the sensitive attribute to one of them according to a given strategy. For this, we propose a strategy to be used in real-life situations and formally prove that there is a higher bound for the expected privacy loss which is sufficiently low.

SA Framework based De-anonymization of Social Networks

The data of social networks contains a large amount of personal information, of which may be public or insignificant, but some may be sensitive and private. Once the user privacy leaked, it may bring a variety of troubles for users. Holders of social networking data first conduct anonymization before the data is published. However, the simple anonymity method does not play a very good protection. At present, a number of de-anonymization attacks for the data releasing of social networks have arisen. These de-anonymization attacks are mostly based on the network structure with the use of feature matching and other methods. In this paper, we model the social network as a Structure-Attribute (SA) framework which integrates the structural characteristics of social networks and social node properties, adding some attribute nodes to the social network. The similarity measurement of social network nodes is proposed, with the consideration of the structure similarity and attribute similarity. The accuracy of node matching and de-anonymization is greatly improved. We conduct our de-anonymization based on the realistic dataset to verify the accuracy and efficiency.

Deeply Understanding Structure-based Social Network De-anonymization

Anonymization techniques are widely adopted to protect users’ privacy during social data publishing and sharing. In this paper, we conduct a comprehensive analysis on the typical structure-based social network de-anonymization algorithms. We aim to understand the de-anonymization approaches and disclose the impacts on their application performance caused by different factors. We design the analysis framework and define three experiment environments to evaluate a few factors’ impacts on the target algorithms. Based on our analysis architecture, we simulate two typical de-anonymization algorithms and evaluate their performance under different pre-configured environments.

De-Anonymizing Social Networks With Random Forest Classifier

Personal privacy is facing severe threats as social networks are sharing user data with advertisers, application developers, and data mining researchers. Although these data are anonymized by removing personal information, such as user identity, nickname, or address information, personal information still could not be protected effectively. In order to arouse the attention of people from academia and industry for privacy protection, we propose a random forest method to de-anonymize social networks. First, we convert the social network de-anonymization problem into a binary classification problem between node pairs. In order to partition large sparse social networks, we use the spectral partition method to partition large graphs into a number of small subgraphs. Then, we use the features of the network structure to train the random forest classifier. As a result, candidate node pairs from anonymous network and auxiliary network can be classified as matched pair by the random forest classifier. Furthermore, we improve the efficiency of our solution through parallelizing proposed method. The experiments conducted on the real data sets show that our solution’s area under the curve is 19% higher than baseline methods on average. Besides that we test the robustness of the proposed algorithm by adding some noisy data, and the result demonstrates that our solution has good robustness.

Analysis of Grasshopper, a Novel Social Network De-anonymization Algorithm

Social networks have an important and possibly key role in our society today. In addition to the benefits, serious privacy concerns also emerge: there are algorithms called de-anonymization attacks that are capable of re-identifying large fractions of anonymously published networks. A strong class of these attacks solely use the network structure to achieve their goals. In this paper we propose a novel structural de-anonymization attack called Grasshopper. By measurements we compare Grasshopper to the state-of-the-art algorithm, and highlight its enhanced capabilities, such as having negligible error rates and accessing yield levels that was not possible before: in cases when there is greater noise in the background knowledge. We furthermore evaluate an anonymity measure for the Grasshopper algorithm which enables the approximate ranking of nodes according to their re-identification rates. Finally, we characterize the robustness of Grasshopper in tackling identity separation, a privacy-enhancing technique that facilitate hiding of structural information.