Privacy-preserving Publishing Research Articles

PanDa Game: Optimized Privacy-Preserving Publishing of Individual-Level Pandemic Data Based on a Game Theoretic Model.

Sharing individual-level pandemic data is essential for accelerating the understanding of a disease. For example, COVID-19 data have been widely collected to support public health surveillance and research. In the United States, these data are typically de-identified before publication to protect the privacy of the corresponding individuals. However, current data publishing approaches for this type of data, such as those adopted by the U.S. Centers for Disease Control and Prevention (CDC), have not flexed over time to account for the dynamic nature of infection rates. Thus, the policies generated by these strategies have the potential to both raise privacy risks or overprotect the data and impair the data utility (or usability). To optimize the tradeoff between privacy risk and data utility, we introduce a game theoretic model that adaptively generates policies for the publication of individual-level COVID-19 data according to infection dynamics. We model the data publishing process as a two-player Stackelberg game between a data publisher and a data recipient and then search for the best strategy for the publisher. In this game, we consider 1) average performance of predicting future case counts; and 2) mutual information between the original data and the released data. We use COVID-19 case data from Vanderbilt University Medical Center from March 2020 to December 2021 to demonstrate the effectiveness of the new model. The results indicate that the game theoretic model outperforms all state-of-the-art baseline approaches, including those adopted by CDC, while maintaining low privacy risk. We further perform an extensive sensitivity analyses to show that our findings are robust to order-of-magnitude parameter fluctuations.

Distributed GAN-Based Privacy-Preserving Publication of Vertically-Partitioned Data

In the era of big data, user data are often vertically partitioned and stored at different local parties. Exploring the data from all the local parties would enable data analysts to gain a better understanding of the user population from different perspectives. However, the publication of vertically-partitioned data faces a dilemma: on the one hand, the original data cannot be directly shared by local parties due to privacy concerns; on the other hand, independently privatizing the local datasets before publishing may break the potential correlation between the cross-party attributes and lead to a significant utility loss. Prior solutions compute the privatized multivariate distributions of different attribute sets for constructing a synthetic integrated dataset. However, these algorithms are only applicable for low-dimensional structured data and may suffer from large utility loss with the increase in data dimensionality. Following the idea of synthetic data generation, we propose VertiGAN, the first framework based on a generative adversarial network (GAN) for publishing vertically-partitioned data with privacy protection. The framework adopts a GAN model comprised of one multi-output global generator and multiple local discriminators. The generator is collaboratively trained by the server and local parties to learn the distribution of all parties' local data and is used to generate a high-utility synthetic integrated dataset on the server side. Additionally, we apply differential privacy (DP) during the training process to ensure strict privacy guarantees for the local data. We evaluate the framework's performance on a number of real-world datasets containing 68--1501 classification attributes and show that our framework is more capable of capturing joint distributions and cross-attribute correlations compared to statistics-based baseline algorithms. Moreover, with a privacy guarantee of epsilon=8, our framework achieves around a 2%~15% improvement in classification accuracy compared to the baseline algorithms. Extensive experimental results demonstrate the capability and efficiency of our framework in synthesizing vertically-partitioned data while striking a satisfactory utility-privacy balance.

ZoomP3: Privacy-Preserving Publishing of Online Video Conference Recordings

The COVID-19 epidemic has made online video conferencing extremely popular throughout the world, with many schools, companies and government sectors using video conferencing applications (e.g., Zoom, Google Meet) in a daily basis. These applications also provide local or cloud recording services, which allow the replay or sharing of video conference recordings (VCRs) in a later time. Such convenience, however, can easily cause infringement of privacy as meeting participants’ personally identifiable information (e.g., face, name, voice) may be exposed to the public without their awareness or consent. While privacy regulation and training can help relieve the situation, efficient and effective tools are also highly desired to protect the privacysensitive users in the VCRs before their public releases. In this work, we propose the first Privacy-Preserving Publishing system (ZoomP3 ) that automatically processes video and audio information in VCRs for privacy protection. Besides leveraging and integrating multiple state-of-the-art computer vision and audio processing tools seamlessly into our system, a number of optimization algorithms are proposed to improve the scalability of the system, enabling it to protect the privacy of long video conferences. We have conducted various tests with short and long videos, and the results (with online demos) verified that ZoomP3 system is suitable for largescale use. It may be applied as an online service, e.g., by Zoom, or by large organizations such as universities, research institutes and government sectors.

Rank-Label Anonymization for the Privacy-Preserving Publication of a Hypergraph Structure

Social networks are often published in the form of a simple graph. The simple graph representation of a social graph shows the dyadic relationship among the social entities whereas it is unable to efficiently represent the relationship among more than two entities, such as the relationship found in the social groups. This type of relationship is called super-dyadic relationship, and it can be effectively represented by a hypergraph model. This work proposes an anonymization scheme called rank-label anonymization for the privacy-preserving publication of a hypergraph structure. Here, an attack model called rank-label attack is proposed, and an anonymization solution is provided to counter this attack. The percentage of disclosure risk shows that the rank-label attack is stronger than the existing rank attack. We propose a method based on sequential clustering to achieve rank-label anonymization called sequential rank-label anonymization (SA). Another algorithm called greedy rank-label anonymization (GA) is also proposed. The quality of the anonymization solution reported by SA and GA is compared with the help of normalized anonymization cost ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">NCost</i> ). Results show that the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">NCost</i> reported by SA is less than that of GA for both Adult and MAG-10 datasets. In Adult dataset, approximately 58% and 62% reduction in the average execution time of GA and SA are obtained than that of a general-purpose computing system due to the use of a high-performance computing system. In MAG-10 dataset, this average reduction percentage is reported to be 56% for GA and 53% for SA. The time complexity of SA is found to be O(n <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sup> ) whereas it is O(n <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> ) in case of GA.

Personalized Privacy-Preserving Publication of Trajectory Data by Generalization and Distortion of Moving Points

Personalized Privacy-Preserving Publication of Trajectory Data by Generalization and Distortion of Moving Points

A Privacy-Preserving Trajectory Publication Method Based on Secure Start-Points and End-Points

By judging whether the start-point and end-point of a trajectory conform to the user’s behavioral habits, an attacker who possesses background knowledge can breach the anonymous trajectory. Traditional trajectory privacy preservation schemes often generate an anonymous set of trajectories without considering the security of the trajectory start- and end-points. To address this problem, this paper proposes a privacy-preserving trajectory publication method based on generating secure start- and end-points. First, a candidate set containing a secure start-point and end-point is generated according to the user’s habits. Second, k−1 anonymous trajectories are generated bidirectionally according to that secure candidate set. Finally, accessibility corrections are made for each anonymous trajectory. This method integrates features such as local geographic reachability and trajectory similarity when generating an anonymized set of trajectories. This provides users with privacy preservation at the k-anonymity level, without relying on the trusted third parties and with low algorithm complexity. Compared with existing methods such as trajectory rotation and unidirectional generation, theoretical analysis and experimental results on the datasets of real trajectories show that the anonymous trajectories generated by the proposed method can ensure the security of trajectory privacy while maintaining a higher trajectory similarity.

Effective Removal of Privacy Breaches in Disassociated Transactional Datasets

A broad range of web activities such as querying web pages, e-commerce transactions, health diagnosis and seat reservations generate vast volume of data, referred to as transactional data. These transactional data are published and widely used for data mining, research and analysis. However, the publishing of individuals’ transactional data implies serious concerns related to privacy for the individuals whose data have been published. The methods proposed in previous researches to preserve the privacy are suitable for structured relational data but are not well suitable to anonymize transactional data since the latter are generally unstructured, sparse and high dimensional. This paper addresses the problem of privacy-preserving publication of transactional data using two enhanced versions of ‘disassociation’ technique. Disassociation limits privacy breaches and increases the utility of the published data, but still, it does not eliminate them because it results in a cover problem that may lead to further privacy concerns. In this paper, we propose two algorithms: (i) improvement in disassociation using suppression and addition (IDSA) and (ii) improvement in disassociation by generalizing cover item (IDGC) to eliminate the cover problem of disassociated data. The proposed algorithms are implemented on INFORMS and BMS-Webview1 datasets and compared to disassociation concerning prevention of privacy breaches as well as loss in information. The results depict that the IDSA leads to a significant drop in privacy breaches due to cover problem with minimal information loss and IDGC completely removes the privacy breaches due to cover problem without any significant loss in data utility.

Robust active attacks on social graphs

In order to prevent the disclosure of privacy-sensitive data, such as names and relations between users, social network graphs have to be anonymised before publication. Naive anonymisation of social network graphs often consists in deleting all identifying information of the users, while maintaining the original graph structure. Various types of attacks on naively anonymised graphs have been developed. Active attacks form a special type of such privacy attacks, in which the adversary enrols a number of fake users, often called sybils, to the social network, allowing the adversary to create unique structural patterns later used to re-identify the sybil nodes and other users after anonymisation. Several studies have shown that adding a small amount of noise to the published graph already suffices to mitigate such active attacks. Consequently, active attacks have been dubbed a negligible threat to privacy-preserving social graph publication. In this paper, we argue that these studies unveil shortcomings of specific attacks, rather than inherent problems of active attacks as a general strategy. In order to support this claim, we develop the notion of a robust active attack, which is an active attack that is resilient to small perturbations of the social network graph. We formulate the design of robust active attacks as an optimisation problem and we give definitions of robustness for different stages of the active attack strategy. Moreover, we introduce various heuristics to achieve these notions of robustness and experimentally show that the new robust attacks are considerably more resilient than the original ones, while remaining at the same level of feasibility.

Restricted Sensitive Attributes-based Sequential Anonymization (RSA-SA) approach for privacy-preserving data stream publishing

Data streams have become a widely-adopted data representation format in many real-world applications. This data streaming may be published for different scientific research, mining, or analysis purposes. However, such streams may contain personal-specific data that could be considered as sensitive about individuals. This makes the privacy preserving of data streams against privacy disclosure attacks, while maintaining their utilization, is a real challenge. Some studies have considered privacy-preserving publishing over data streams with only Single Sensitive Attribute, in which they do not protect the published streams from all possible privacy attacks. In this paper, we propose a novel Restricted Sensitive Attributes-based Sequential Anonymization (RSA-SA) approach for privacy-preserving data stream publishing. Besides, two new privacy restrictions are introduced to restrict the published Sensitive Attributes values: Semantic-diversity and Sensitivity-diversity. RSA-SA can protect the sensitive values of the published data streams against the related privacy attacks, including the attribute disclosure, skewness, similarity, and sensitivity attacks. In addition, RSA-SA handles data streams that have either single or multiple sensitive attributes with minimum information loss and delay time. Thus, the data utility of the published data streams is efficiently maintained to provide more accurate mining and analytical results, where robust invulnerability to privacy attacks is sustained.

A relative privacy model for effective privacy preservation in transactional data

SummaryData publishing is pivotal to advances in knowledge discovery. Nonetheless, such publishing may suffer from privacy disclosures. This is especially the case in transactional data such as web search and point of sales logs. The reason is that the current potent privacy preserving mechanisms mainly focus on relational data. In this work, we propose a new privacy metric for transactional data to prevent inference attacks by ensuring that the adversary learns no more about an intended victim than what is publicly available. We then propose a publication mechanism Anony, which satisfies our privacy metric without excessive loss of utility. Finally, we present an empirical evaluation of our method on three benchmark datasets, and the results show the effectiveness of our method.

Release Connection Fingerprints in Social Networks Using Personalized Diffierential Privacy

In social networks, different users may have different privacy preferences and there are many users with public identities. Most work on differentially private social network data publication neglects this fact. We aim to release the number of public users that a private user connects to within n hops, called n-range Connection fingerprints(CFPs), under user-level personalized privacy preferences. We proposed two schemes, Distance-based exponential budget absorption (DEBA) and Distance-based uniformly budget absorption using Ladder function (DUBA-LF), for privacy-preserving publication of the CFPs based on Personalized differential privacy(PDP), and we conducted a theoretical analysis of the privacy guarantees provided within the proposed schemes. The implementation showed that the proposed schemes are superior in publication errors on real datasets.

Achieving Optimal K-Anonymity Parameters for Big Data

Datasets containing private and sensitive information are useful for data analytics. Data owners cautiously release such sensitive data using privacy-preserving publishing techniques. Personal re-identification possibility is much larger than ever before. For instance, social media has dramatically increased the exposure to privacy violation. One well-known technique of k-anonymity proposes a protection approach against privacy exposure. K-anonymity tends to find k equivalent number of data records. The chosen attributes are known as Quasi-identifiers. This approach may reduce the personal re-identification. However, this may lessen the usefulness of information gained. The value of k should be carefully determined, to compromise both security and information gained. Unfortunately, there is no any standard procedure to define the value of k. The problem of the optimal k-anonymization is NP-hard. In this paper, we propose a greedy-based heuristic approach that provides an optimal value for k. The approach evaluates the empirical risk concerning our Sensitivity-Based Anonymization method. Our approach is derived from the fine-grained access and business role anonymization for big data, which forms our framework.

Towards Privacy Preserving Publishing of Set-Valued Data on Hybrid Cloud

Storage as a service has become an important paradigm in cloud computing for its great flexibility and economic savings. However, the development is hampered by data privacy concerns: data owners no longer physically possess the storage of their data. In this work, we study the issue of privacy-preserving set-valued data publishing. Existing data privacy-preserving techniques (such as encryption, suppression, generalization) are not applicable in many real scenes, since they would incur large overhead for data query or high information loss. Motivated by this observation, we present a suite of new techniques that make privacy-aware set-valued data publishing feasible on hybrid cloud. On data publishing phase, we propose a data partition technique, named extended quasi-identifier-partitioning (EQI-partitioning), which disassociates record terms that participate in identifying combinations. This way the cloud server cannot associate with high probability a record with rare term combinations. We prove the privacy guarantee of our mechanism. On data querying phase, we adopt interactive differential privacy strategy to resist privacy breaches from statistical queries. We finally evaluate its performance using real-life data sets on our cloud test-bed. Our extensive experiments demonstrate the validity and practicality of the proposed scheme.

Privacy-Preserving Publishing of Multilevel Utility-Controlled Graph Datasets

Conventional private data publication schemes are targeted at publication of sensitive datasets either after the k -anonymization process or through differential privacy constraints. Typically these schemes are designed with the objective of retaining as much utility as possible for the aggregate queries while ensuring the privacy of the individual records. Such an approach, though suitable for publishing aggregate information as public datasets, is inapplicable when users have different levels of access to the same data. We argue that existing schemes either result in increased disclosure of private information or lead to reduced utility when some users have more access privileges than the others. In this article, we present an anonymization framework for publishing large datasets with the goals of providing different levels of utility to the users based on their access privilege levels. We design and implement our proposed multilevel utility-controlled anonymization schemes in the context of large association graphs considering three levels of user utility, namely, (1) users having access to only the graph structure, (2) users having access to the graph structure and aggregate query results, and (3) users having access to the graph structure, aggregate query results, and individual associations. Our experiments on real large association graphs show that the proposed techniques are effective and scalable and yield the required level of privacy and utility for each user privacy and access privilege level.

Local Suppression and Splitting Techniques for Privacy Preserving Publication of Trajectories

We study the problem of preserving user privacy in the publication of location sequences. Consider a database of trajectories, corresponding to movements of people, captured by their transactions when they use credit cards, RFID debit cards, or NFC ( http://en.wikipedia.org/wiki/Near_field_communication ) compliant devices. We show that, if such trajectories are published exactly (by only hiding the identities of persons that followed them), one can use partial trajectory knowledge as a quasi-identifier for the remaining locations in the sequence. We devise four intuitive techniques, based on combinations of locations suppression and trajectories splitting, and we show that they can prevent privacy breaches while keeping published data accurate for aggregate query answering and frequent subsets data mining.

Privacy-Preserving Publishing of Hierarchical Data

Many applications today rely on storage and management of semi-structured information, for example, XML databases and document-oriented databases. These data often have to be shared with untrusted third parties, which makes individuals’ privacy a fundamental problem. In this article, we propose anonymization techniques for privacy-preserving publishing of hierarchical data. We show that the problem of anonymizing hierarchical data poses unique challenges that cannot be readily solved by existing mechanisms. We extend two standards for privacy protection in tabular data ( k -anonymity and ℓ-diversity) and apply them to hierarchical data. We present utility-aware algorithms that enforce these definitions of privacy using generalizations and suppressions of data values. To evaluate our algorithms and their heuristics, we experiment on synthetic and real datasets obtained from two universities. Our experiments show that we significantly outperform related methods that provide comparable privacy guarantees.

DPcode: Privacy-Preserving Frequent Visual Patterns Publication on Cloud

Nowadays, cloud has become a promising multimedia data processing and sharing platform. Many institutes and companies plan to outsource and share their large-scale video and image datasets on cloud for scientific research and public interest. Among various video applications, the discovery of frequent visual patterns over graphical data is an exploratory and important technique. However, the privacy concerns over the leakage of sensitive information contained in the videos/images impedes the further implementation. Although the frequent visual patterns mining (FVPM) algorithm aggregates summary over individual frames and seems not to pose privacy threat, the private information contained in individual frames still may be leaked from the statistical result. In this paper, we study the problem of privacy-preserving publishing of graphical data FVPM on cloud. We propose the first differentially private frequent visual patterns mining algorithm for graphical data, named DPcode. We propose a novel mechanism that integrates the privacy-preserving visual word conversion with the differentially private mechanism under the noise allocation strategy of the sparse vector technique. The optimized algorithms properly allocate the privacy budgets among different phases in FPM algorithm over images and reduce the corresponding data distortion. Extensive experiments are conducted based on datasets commonly used in visual mining algorithms. The results show that our approach achieves high utility while satisfying a practical privacy requirement.

Anonymization in Social Ne tworks: A Survey on the issues of Data Privacy in Social Network Sites

Recent years have seen unprecedented growth in the popularity of social network systems. Social networks are online applications which allow its users to connect by various link types. Online social networks, such as Facebook, LinkedIn are progressively utilized by many people. It makes digital communication technologies sharpening tools for extending the social circle of people. It has already become an important integral part of our daily lives, enabling us to contact our friends and families on time. As social networks have developed rapidly, recent research has begun to explore social networks to understand their structure, advertising and marketing, and data mining online population, representing 1.2 billion users around the world. More and more social network data has been made publicly available and analyzed in one way or another. But some of the information revealed is meant to be private hence social network data has led to the risk of leakage of confidential information of individuals. However, privacy concerns can be used to prevent these efforts. Privacy pre- serving publishing of social network data becomes a more and more important concern. This leads for privacy-preserving social network data mining, which is the discovery of information and relationships from social network data without violating privacy. Privacy in online social networks data has been of utmost concern. Hence, the research in this field is still in its early years. Privacy preservation is a significant research issue in social networking. Since more personalized information is shared with the public, violating the privacy of a target user become much easier. We argue that the different privacy problems are entangled and that research on privacy in OSNs would benefit from a more holistic approach. In this paper, after an overview on different research subareas of SNSs, we will get more focused on the subarea of privacy protection in social network and we present a brief yet systematic review of the existing anonymization techniques for privacy preserving publishing of social net- work data.

Logical Foundations of Privacy-Preserving Publishing of Linked Data

The widespread adoption of Linked Data has been driven by the increasing demand for information exchange between organisations, as well as by data publishing regulations in domains such as health care and governance. In this setting, sensitive information is at risk of disclosure since published data can be linked with arbitrary external data sources. In this paper we lay the foundations of privacy-preserving data publishing (PPDP) in the context of Linked Data. We consider anonymisations of RDF graphs (and, more generally, relational datasets with labelled nulls) and define notions of safe and optimal anonymisations. Safety ensures that the anonymised data can be published with provable protection guarantees against linking attacks, whereas optimality ensures that it preserves as much information from the original data as possible, while satisfying the safety requirement. We establish the complexity of the underpinning decision problems both under open-world semantics inherent to RDF and a closed-world semantics, where we assume that an attacker has complete knowledge over some part of the original data.

Utility-based anonymisation for dataset with multiple sensitive attributes

Privacy-preserving data publication problem has attracted more and more attentions in recent years. A lot of related research works have been done towards dataset with single sensitive attribute. However, usually, original dataset contains more than one sensitive attribute. In this paper, we apply k-anonymity principle to solve the data publication problem for dataset with multiple sensitive attributes. We first cluster sensitive values based on a utility matrix. Then, we use a greedy strategy to partition tuples into equivalence classes. Our method can guarantee that the size of equivalence class is k except the last one, which reduces information loss. Also, we can guarantee the diversity of sensitive value in an equivalence class, which can protect privacy against the homogeneity attack. Experiments on a real dataset show that our method performs well on information loss, which indicates that we can guarantee data utility while protecting personal privacy.