Privacy Leakage and the Manipulation of Public Opinion in Online Social Networks

Luceri, Luca (2020). Privacy Leakage and the Manipulation of Public Opinion in Online Social Networks. (Dissertation, Institute of Computer Science, Faculty of Science)

Preview

Text phd_thesis-36.pdf - Published Version Available under License Creative Commons: Attribution-Noncommercial-No Derivative Works (CC-BY-NC-ND). Download (7MB) | Preview

Online Social Networks (OSNs) are computer-based technologies that enable users to create content, share information, and establish social relationships in online platforms. The advent of OSNs has dramatically revolutionized the way we access the news, share opinion, make business and politics. Although the wide adoption of OSNs brought several positive effects, the combination of its technological and social aspects hides harmful effects for both the individual users and the entire society. Among the potential risks analyzed in the literature (e.g., security, health, etc.), in this thesis, we analyze the perils related to the privacy leakage and the manipulation of opinions in OSNs. In particular, we investigate the factors driving these perils, with the final objective of raising users’ awareness of the risks behind their online activities. We show how, for both the privacy and manipulation perils, social connections play a central role in fostering and exacerbating such issues. In fact, social connections among OSN users result in a network structure, which enables the spreading of information, behaviours, and opinions across the OSN population through online interactions. Along this research direction, we first explore to what extent an individual’s privacy can be violated by leveraging information provided by other users in the OSN. In particular, we examine the problem of location privacy by developing methods to assess users’ privacy risks and strategies to control the public exposure of their data. Then, we explore the privacy peril by considering the diffusion of behaviours and opinions in OSNs. In fact, social interactions can substantially affect the extent to which a behaviour, an opinion, or a product is adopted by OSN users. This concept is a social phenomenon referred to as social influence. According to this concept, we investigate whether social influence modelling (i.e., learning influence strengths among subjects) can be used to accurately predict users’ future activity and, in turn, violate their privacy. We present different approaches to model social influence and we show how such models can be employed to violate users’ privacy. Online interactions and social influence play also a crucial role in the manipulation of peoples’ belief and opinion. Manipulation campaigns have raised particular concerns in the political context. Bots (i.e., software-controlled accounts) and trolls (i.e., state-sponsored human operators) are the main actors responsible for these campaigns. In this thesis, we analyze the activity of such malicious actors to enhance and enable countermeasures for their detection. More specifically, we first uncover the strategies employed by bots to avoid detection and manipulate human users. Then, we present an approach for detecting trolls’ activity in OSNs that accurately identifies troll accounts and unveils their distinguishing behaviour with respect to regular users. The results presented in this thesis confirm the privacy and manipulation risks in OSNs: On one hand, we prove that users’ privacy is not under individual control as public information can be efficiently used to predict their behaviour, and in turn, violate their privacy. On the other hand, we show that malicious actors have become increasingly sophisticated to escape detection andmanipulate human users. However, the majority of OSN users are not conscious or underestimate the potential risks behind their online activity. Towards raising users’ awareness of such perils and to mitigate this set of open problems, we propose an awareness service, based on a mobile application, to timely communicate users their current risks in OSNs. For this purpose, we deploy a framework to collect users’ data in a privacy-preserving way and provide them with feedback about their privacy and manipulation risks in real-time.

Interest & Impact

Downloads

159 since deposited on 03 Apr 2020
27 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item