A semantic context model for managing privacy on smartphones

In this work we describe an approach to protecting the privacy and security of user data on mobile devices using a richer semantic model of a user's context. Mobile OS frameworks like Android always lacked mechanisms for dynamic privacy control, while recent advances in context modeling, tracking and collaborative localization has led to the emergence of a new class of smartphone applications that can access and share embedded sensor data. Existing literature on context based privacy and security has predominantly focused on device user's context for probing privacy vulnerability and enforcing security at runtime. We bring into picture the most important component of privacy vulnerability on smartphones, the resident applications themselves and we introduce the novel idea of application provenance. Our context model is realized as a dynamic knowledge base of RDF [67] triples grounded in an ontology in the semantic web language OWL. Policies in the form of rules over this knowledge base monitor and control application access to sensitive information and sensor data. The policies filter data flowing from sensor resources to applications to reduce disclosure by generalizing or obfuscating data. Our ontology includes the ability to represent application provenance and other metadata that can be used by the policies. The resulting system provides fine-grained, context-dependent control to sensitive user data.