A Sensing Architecture for Empathetic Data Systems

Today's increasingly large and complex databases require novel and machine aided ways of exploring data. To optimize the selection and presentation of data, we suggest an unconventional approach. Instead of exclusively relying on explicit user input to specify relevant information or to navigate through a data space, we exploit the power and potential of the users' unconscious processes in addition. To this end, the user is immersed in a mixed reality environment while his bodily reactions are captured using unobtrusive wearable devices. The users' reactions are analyzed in real-time and mapped onto higher-level psychological states, such as surprise or boredom, in order to trigger appropriate system responses that direct the users' attention to areas of potential interest in the visualizations. The realization of such a close experience-based human-machine loop raises a number of technical challenges, such as the real-time interpretation of psychological user states. The paper at hand describes a sensing architecture for empathetic data systems that has been developed as part of such a loop and how it tackles the diverse challenges.