Privacy-Preserving Ubiquitous Social Mining via Modular and Compositional Virtual Sensors

The introduction of ubiquitous systems, wearable computing and ´Internet of Things´ technologies in our digital society results in a large-scale data generation. Environmental, home, and mobile sensors are only a few examples of the significant capabilities to collect massive data in real-time from a plethora of heterogeneous social environments. These capabilities provide us with a unique opportunity to understand and tackle complex problems with new novel approaches based on reasoning about data. However, existing ´Big Data´ approaches often turn this opportunity into a threat of citizens´ privacy and open participation by surveilling, profiling and discriminating people via closed proprietary data mining services. This paper illustrates how to design and build an open participatory platform for privacy-preserving social mining: the Planetary Nervous System. Building such a complex platform in which data sharing and collection is self-determined by the user and is performed in a decentralized fashion within different ubiquitous environments is a challenge. This paper tackles this challenge by introducing a modular and compositional design approach based on a model of virtual sensors. Virtual sensors provide a holistic approach to build the core functionality of the Planetary Nervous System but also social mining applications that extend the core functionality. The holistic modeling approach with virtual sensors has the potential to simplify the engagement of citizens in different innovative crowd-sourcing activities and increase its adoption by building communities. Performance evaluations of virtual sensors in the Planetary Nervous System confirm the feasibility of the model to build real-time ubiquitous social mining services.