Knowledge Discovery using Geosensor Networks

Geosensor technology is radically changing our ability to collect data in a wide range of domains. While sensor hardware and low-level system technology is advancing rapidly, higher level modeling needs to be advanced in parallel, so that users can effectively utilize the potential. This paper proposes an approach to modeling interaction with geosensor networks based on dynamic primitives. Much of the current spatiotemporal information system technology is based on the snapshot metaphor, where the dynamic world is imagined as a temporal sequence of snapshots. This imposes limitations on the representations and reasoning capabilities of such systems. In particular, it is not possible using temporal snapshots to explicitly model events, processes, and actions. This paper summarizes some of the work done on extensions to current models to provide fuller event modeling capabilities, and gives examples of applications to the geosensor and geographic domain. The work will have impact in understanding physical aspects of the global security environment, because event-based models are needed for effective and high-level handling of geospatial intelligence information.