Strategies for Privacy Negotiation in Online Social Networks

Dilara Keküllüoglu, Nadin Kökciyan, Pinar Yolum

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

Online social networks are changing the way information is shared among individuals. Contrary to traditional Web systems, such as e-commerce Web sites, where information about a user is managed solely by the user herself, in online social networks other users can contribute to the content that is shared about an individual. The shared content may reveal information about the user, which the user might not wanted to share herself. This creates a privacy breach on the user’s side. In current online social networks, a common way to deal with this is for the user to complain to the social network administration and ask the content to be removed. However, by the time the content is removed (if at all), many people might have seen it already. Ideally, it would be best if such a content was not shared in the first place.
Recent work on privacy management has focused on applying agreement technologies to solve privacy problems before they take place. Two important works in this line are that of Mester et al. and Such and Rovatsos. These approaches apply negotiation techniques to resolve privacy conflicts among users. They both consider negotiation before a content is being shared. Both approaches assume that negotiation is being performed on a single content and cannot account for ongoing interactions. However, it has been observed that users build reciprocal trust in online social networks and respect others as much as others respect them. Hence, it is of utmost importance to consider repeated interactions, as opposed to single interactions, to study privacy leakages.
This paper proposes a multiagent management of privacy in online social networks, where each user is represented by an agent that helps its user preserve its privacy. The privacy of users is preserved by a hybrid negotiation architecture where privacy domain and rules are represented semantically but the decision making is done by the agents using utility functions. The paper develops various negotiation strategies including one that exploits reciprocity. The key idea is that each agent keeps track of whether a certain other user has been helpful before in preserving privacy using a credit system. When agents help others in preserving their privacy, their credit increases so that later they can ask others to help them. Hence, helping others to preserve privacy serves as an incentive. Using these strategies, agents can negotiate on the content and agree on how it will be shared before the post goes online.
Original languageEnglish
Title of host publicationEuropean Conference on Artificial Intelligence (ECAI)
EditorsGal A. Kaminka, Maria Fox, Paolo Bouquet, Eyke Hüllermeier, Virginia Dignum, Frank Dignum, Frank van Harmelen
PublisherIOS Press
Number of pages2
ISBN (Electronic)978-1-61499-672-9
ISBN (Print)978-1-61499-671-2
Publication statusPublished - 2 Sept 2016

Publication series

NameFrontiers in Artificial Intelligence and Applications
PublisherIOS Press
ISSN (Print)0922-6389
ISSN (Electronic)1879-8314


Dive into the research topics of 'Strategies for Privacy Negotiation in Online Social Networks'. Together they form a unique fingerprint.

Cite this