Contextual Integrity for Argumentation-based Privacy Reasoning

Privacy management in online systems is a complex task. Recently, contextual integrity theory has been introduced to model privacy, which considers the social contexts of users before making privacy decisions. However, having a practical application based on this theory is not straightforward. In this paper, we propose an agent-based framework for privacy policy reasoning that combines the power of ontologies together with argumentation techniques to resolve privacy conflicts. First, we propose an ontology that represents the contextual integrity theory. We then introduce an argumentation-based dialogue framework that could: (i) reason about contextual norms to resolve privacy conflicts among agents, and (ii) provide justifications to the agents during multi-party dialogues. We apply our approach to privacy scenarios in various contexts where each scenario has different challenges to address. We conclude with theoretical results to show the effectiveness of the framework.