Optimal tag suppression for privacy protection in the semantic Web

et al. 2016

Knowl Inf Syst

Self Cite

Abstract-Despite the several advantages commonly attributed to social networks such as easiness and immediacy to communicate with acquaintances and friends, significant privacy threats provoked by unexperienced or even irresponsible users recklessly publishing sensitive material are also noticeable. Yet, a different, but equally significant privacy risk might arise from social networks profiling the online activity of their users based on the timestamp of the interactions between the former and the latter. In order to thwart this last type of commonly neglected attacks, this paper proposes an optimized deferral mechanism for messages in online social networks. Such solution suggests intelligently delaying certain messages posted by end users in social networks in a way that the observed online-activity profile generated by the attacker does not reveal any time-based sensitive information, while preserving the usability of the system. Experimental results as well as a proposed architecture implementing this approach demonstrate the suitability and feasibility of our mechanism.

Section: Adversary Modelmentioning

confidence: 98%

Section: Privacy-enhancing Technologies Against Time-based Profiling mentioning

confidence: 99%

Shall I post this now? Optimized, delay-based privacy protection in social networks

Parra-Arnau

Mármol²,

et al. 2016

Knowl Inf Syst

Self Cite

“…This is analogous to the histogram of the relative frequency of the different search categories, in the case of the web search presented in [38,52]. This profile reveals information related to different locations, independently of the rest of the visited locations and correlations among them.…”

Section: Location Profilementioning

confidence: 99%

Entropy-Based Privacy against Profiling of User Mobility

Rodríguez-Carrión

Forné

et al. 2015

Entropy

Self Cite

Location-based services (LBSs) flood mobile phones nowadays, but their use poses an evident privacy risk. The locations accompanying the LBS queries can be exploited by the LBS provider to build the user profile of visited locations, which might disclose sensitive data, such as work or home locations. The classic concept of entropy is widely used to evaluate privacy in these scenarios, where the information is represented as a sequence of independent samples of categorized data. However, since the LBS queries might be sent very frequently, location profiles can be improved by adding temporal dependencies, thus becoming mobility profiles, where location samples are not independent anymore and might disclose the user's mobility patterns. Since the time dimension is factored in, the classic entropy concept falls short of evaluating the real privacy level, which depends also on the time component. Therefore, we propose to extend the entropy-based privacy metric to the use of the entropy rate to evaluate mobility profiles. Then, two perturbative mechanisms are considered to preserve locations and mobility profiles under gradual utility constraints. We further use the proposed privacy metric and compare it to classic ones to evaluate both synthetic and real mobility profiles when the perturbative methods proposed are applied. The results prove the usefulness of the proposed metric for mobility profiles and the need Entropy 2015, 17 3914 for tailoring the perturbative methods to the features of mobility profiles in order to improve privacy without completely loosing utility.

“…An example of this latter kind of data perturbation is the elimination of tags as a privacy-enhancing strategy [13,14]. This strategy allows users to preserve their privacy to a certain degree, but it comes at the cost of some degradation in the usability of the service.…”

Section: State-of-the-artmentioning

confidence: 99%

On the Anonymity Risk of Time-Varying User Profiles

Puglisi

Forné

2017

Entropy

Self Cite

Websites and applications use personalisation services to profile their users, collect their patterns and activities and eventually use this data to provide tailored suggestions. User preferences and social interactions are therefore aggregated and analysed. Every time a user publishes a new post or creates a link with another entity, either another user, or some online resource, new information is added to the user profile. Exposing private data does not only reveal information about single users' preferences, increasing their privacy risk, but can expose more about their network that single actors intended. This mechanism is self-evident in social networks where users receive suggestions based on their friends' activities. We propose an information-theoretic approach to measure the differential update of the anonymity risk of time-varying user profiles. This expresses how privacy is affected when new content is posted and how much third-party services get to know about the users when a new activity is shared. We use actual Facebook data to show how our model can be applied to a real-world scenario.