Privacy issues when sharing sensitive semantic locations in urban settings

Since the trajectory data is easily produced, collected and stored the challenge now is to make it useful for various purposes. To this end many location based services are implemented and new data mining/analysis techniques are proposed for better decision making. Privacy requirements, on the other hand, hinder the development of services and analysis as the location data offers intrusive inferences about people’s sensitive whereabouts. So, this makes the privacy protection an important issue when provisioning location based services and trajectory data publication. In this project, privacy issues at sensitive semantic location hiding in urban settings are studied. This project builds upon our previous collaborative work with the host. The project extended the problem formulation, attack modeling, theoretical analysis and algorithms. Most importantly, (i) optimally generating strongly cloaked regions problem is shown to be NP-Hard, (ii) two use cases of location sharing, sporadic and constant, are identified and related attack models with respective solutions are provided, (iii) a solution to off-line publication of whole trajectories are developed, and (iv) model is extended with additional background information including temporal semantics and frequency. The next step will be experimental evaluation of the new theoretical study.