Building an Information System for Reality Mining Based on Communication Traces

Reality mining (RM), i.e. gathering and analysing data about human behaviour and interaction in the real world, is gaining increasing interest in various disciplines. In this paper, we are discussing RM from a computer science perspective. The objective is to propose a reference architecture supporting the technical issues associated with RM. We argue for a web service based architecture providing the necessary performance, flexibility, extensibility and interoperability required in RM. A prototypical implementation is presented, demonstrating how sensor data from communication and location sensors of a group are being sent to a web service for further analysis. The gathered information can be used, for example, to analyse social networks in real time. In our example application, we calculate scores for each individual group members behavioral stereotype, which provides information for detecting bottlenecks within group communication. This information is then reported back, following the idea of sociometry, which reflects the groups current situation back to the group in order to allow improvements. Feedback and reporting are another important functional requirement in RM. In our prototype, we provide such functionality in software. The implementation as a web service allows easy integration into existing information and communication systems.