Abstract:
Online Social Networks (OSNs) are web-services that enable users to connect with other users, share content and view other users' content. Spread of online social networks brings privacy problems that are not addressed before. Privacy is de ned as the right to conceal certain information from designated people. Users share personal information, photos and videos about themselves as well as their friends. OSNs give management rights of a content to the user who uploads them. This can lead to privacy violations for other users that are related to the content. These users may not want some people to see the content and the uploader may not be aware of this preference. Ideally everyone related to a content should have a say on how it is shared. We propose a hybrid negotiation architecture that helps users to solve privacy violations. Every user is represented by an agent that knows the relations and the privacy concerns of the user. We represent these agents and their relations semantically, also enable usage of utility functions for them to reach decisions. We develop various negotiation strategies as well as a trade-o mechanism that uses reciprocity principal to have negotiations that consider past interactions. We introduce a new evaluation metric for measuring the outcome of the negotiations. We run simulations to compare our negotiation strategies. As a result, our proposed strategies perform better than the existing methods of OSNs.