Integration Of Geosensor Data By Means Of An Event Abstraction Layer Alejandro Llaves Institute for Geoinformatics, University of Muenster Weselerstr. 253, 48151 Muenster, Germany Supervisor: Prof. Dr. Werner Kuhn alejandro.llaves@uni-muenster.de Keywords. Spatiotemporal Modeling, Event Processing, Geospatial Events 1. Motivation With the new generation of Sensor Web Enablement 1 (SWE) [2] the number of possibili- ties to access sensor observations has increased considerably. Nowadays, more and more data providers are making their data sets available in the Web using standard services, e.g. Sensor Observation Service (SOS) [7]. These services are easily integrated in Ser- vice Oriented Architectures (SOA), which allow users to get the sensor data following a polling procedure, i.e. sending a request with some parameters and receiving the data in the service response. Some applications deal with time-sensitive issues which require quick responses to certain occurrences, e.g. environmental change monitoring. Handling all the data pro- vided by sensors in near real-time and obtain valuable information is challenging from a technical and a scientific point of view. Event processing methods provide real-time processing capabilities and event filter- ing on continuous time-series of data. The main problem to integrate event processing into existing architectures is the lack of a common vocabulary for event concepts and their relations [3], which causes interoperability issues between event-driven applica- tions. 2. Background The subjectivity of event categorization was discussed by Allen and Ferguson. They claimed that events do not exist in our real world, but events are the result of the clas- sification of patterns of change by individuals acting as observer agents [1]. This state- ment highlights that two communities may have different conceptual representations of the same occurrence, e.g. a heavy rainfall in Egypt may not be considered as such in Germany. 1 http://www.opengeospatial.org/projects/groups/sensorweb