Abstract
Time series of observations reflect the status of environmental properties. Variations in these properties can be considered as events when they potentially affect the stability of the monitored environment. Organisations dedicated to analyse environmental change use institutionalised descriptions of events to define the observable conditions under which events happen. This also applies to the methods used to classify and model changes in environmental monitoring. The heterogeneity of representations often causes interoperability problems when such communities exchange geospatial information. To enhance interoperability among diverse communities, it is required to develop models that do not restrict the representation of events, but allow integrating different perspectives on changes in the environment. The goal of the Event Abstraction Layer is to facilitate the analysis and integration of geosensor data by inferring events from time series of observations. For the analysis of geosensor data, we use event processing to detect event patterns in time series of observations. Spatio-temporal properties of the event are inferred from the geosensor location and the observation timestamps. For the data integration, we represent event-related information extracted from multiples sources under a common event model. Additionally, domain knowledge is modelled in a multilevel ontology structure.
Acknowledgements
This research was carried out at the Institute for Geoinformatics, University of Muenster and it was funded by the European project ENVISION (FP7-249170). We are thankful for discussions with members of the Muenster Semantic Interoperability Lab (MUSIL) and the International Research Training Group on Semantic Integration of Geospatial Information (IRTG-SIGI).
Notes
1. SOS is the OGC standard specification for sensor data retrieval. More information is available at http://www.opengeospatial.org/standards/sos
2. SPARQL is a W3C recommendation language to query RDF data. More details at http://www.w3.org/TR/rdf-sparql-query/
3. O&M is an OGC standard specification to enable sensor data encoding. The O&M website is available at http://www.opengeospatial.org/standards/om
4. RDF is a data model that allows defining statements in the form of subject-object-predicate triples. More information can be found at http://www.w3.org/RDF/
5. More information about Drools can be found at http://www.jboss.org/drools
6. More information about Prova is available at https://prova.ws
8. Reviews available at http://www.w3.org/2005/Incubator/ssn/XGR-ssn-20110628/#Review_of_Sensor_and_ Observation_ontologies.
9. To differentiate concepts of the different ontologies, we will use the ontology acronym as namespace: dul and ssn.
10. A complete definition of dul:Situation is available at http://www.w3.org/2005/Incubator/ssn/wiki/DUL_ssn
11. More details at http://www.w3.org/2005/Incubator/ssn/wiki/Incubator_Report.
12. The RDF Schema Description Language 1.0 is available at http://www.w3.org/TR/2004/REC-rdf-schema-20040210/.
13. More information at http://parliament.semwebcentral.org/.
14. The source code is available at https://github.com/allaves/EPS.
15. Esper for Java available at http://esper.codehaus.org/about/esper/esper.html
16. Esper’s EPL is a SQL-like language, see http://esper.codehaus.org/esper-4.4.0/doc/reference/en/html_%20single/index.html
18. Documentation for N3 is available at http://www.w3.org/TeamSubmission/n3/
19. Ontology available at http://wsmls.googlecode.com/svn/trunk/local/water/0.6/.
20. Official website available at http://www.rowater.ro/default.aspx.
21. Romanian Waters online GIS available at http://gis2.rowater.ro:8989/SituatieHidrologica.html.
22. Descriptions in Romanian are available in the Romanian Waters’ Emergency Management Regulations http://www.rowater.ro/daprut/Documente%20Repository/Regulament%20%20gestionare%20situatii%20de%20urgenta%20.pdf, CAPITOLUL II, Art. 11, Section (2) B.
23. Two observations (obs1, obs2) produced by the same sensor and ordered in time where the value of obs1 is below the attention threshold and the value of obs2 is above.
24. Deceed’ is a neologism that corresponds to the antonym of ‘exceed’. Two observations (obs1, obs2) produced by the same sensor and ordered in time where the value of obs1 is above the attention threshold and the value of obs2 is below.
25. Two events (e1, e2) related to the same gauging station and ordered in time where e1 is of type attention threshold exceeded and e2 is of type attention threshold deceeded.
26. Two events (e1, e2) related to the same gauging station and ordered in time where e1 is of type flooding threshold exceeded and e2 is of type flooding threshold deceeded.
27. Ontology available at http://wsmls.googlecode.com/svn/trunk/application/EventType/RomanianWaters/.
28. Meters above the Adriatic Sea.
29. The application ontology for Hidroelectrica Romania, namespace HR, is available at http://wsmls.googlecode.com/svn/trunk/application/EventType/IronGates/.
30. The gauging station of Pancevo is located at coordinates 4447′53″N 2038′13″E.
31. The code of the real-time experiment is available at https://github.com/allaves/EPS/blob/master/EventProcessingServiceClient/test/de/ifgi/envision/eps/thesis/Danube/IGDEALThesisRealTimeDataTest.java.
32. Corresponding namespaces must be added above the query with the format ‘PREFIX [namespace]’).
33. The event instances exported from Parliament are available at https://www.dropbox.com/s/gkes69fod08kx1s/realTimeExperiment_parliamentExport.n3.
34. The EPS is running on Ubuntu 12.04.2 LTS, with an Intel Xeon CPU E5530 @ 2,40 GHz processor and a cache size of 8KB. The RAM memory has 1 GB.
35. Tim Berners-Lee defined the Linked Data Principles at http://www.w3.org/DesignIssues/LinkedData.html.