Argumentation for protecting users' privacy in online social networks

Abstract

Preserving privacy in online social networks (OSNs) is becoming increasingly important as more people expose their personal information in those networks. A social network user can easily share a post concerning other users. However, privacy constraints of these relevant users may be di erent from each other so that privacy violations occur. Therefore, all relevant users must be able to engage in a discussion and express themselves, so that they can protect their privacy. We propose PriArg to enable relevant users discuss on a post using Assumption- Based Argumentation (ABA). We present each user in an OSN with an agent, which is equipped with an ontology. Ontologies involve the social network knowledge of users as well as their privacy constraints. Hence, agents are fully aware of the privacy constraints of their user and can act on behalf of them. When an agent wants to share a post, an argumentation session starts between the agent and other relevant agents. They provide each other arguments to express themselves and to convince other agents that their claim is true. At the end of the argumentation, we use an ABA engine to nd whether sharing the post is justi ed according to the provided arguments. Agents may not have necessary information in their ontology to support their arguments. In this case, they can ask other agents in the social network for information. We evaluate our approach by showing its applicability with the real world scenarios. Moreover, we conduct a survey and an interview to compare the PriArg results with user expectations. At last, we compare the related work in the literature with PriArg.