User performs analysis via destop GIS applications (in the experiment were used Quantum GIS and uDig) and reasoning application interface. The experimental phase was used Quantum GIS, connected through the http protocol to PostgreSQL / PostGIS.
To start the analysis is needed, spatial data resource stored in relational database, that contains a set of layers of an SDI. The experiment involved cadastral data, technical infrastructure data, road network data and land use planning data. A set of layers was chosen so that was sufficient to resolve the problem. The information about the needed set of layers is obtained from the domain and decision criteria ontologies. The information about needed resources is provided by reasoning application.
User creates a project in destop GIS application, to which imports the necessary layers. The clear visualization of resources is provided by layers styling. Styling let then interpret the results of the analysis.
For performance reasons, the subject of analysis involves only a part of the area of the imported data. This is done by layer ‘bb’, which always contains only one polygon. Polygon serves the choice of the area of analysis. Bb layer reduces the number of objects to be analyzed in the individual layers. After determining the area of analysis and save it in the database, you can begin the analysis.
In the experiment, we make requests for the characteristics of parcels that meet certain conditions. From ontology of the knowledge base we can choose different classes of parcels, corresponding to different characteristics.
Start-up analysis can be done from the command line on your computer after downloading the reasoning application (eg sdssDecisionResultInferenceWogis2DB.jar). The call has form:
java -Xmx1280M -jar sdssDecisionResultInferenceWogis2DB.jar http://www.semgis.st7.eu/owl/sdss_plankb_tbox.owl ParcelConformingGeneralRequirementsOfSpatPlanActForDevConditionsDec.
The analysis can also be called via http://wogis2.igig.up.wroc.pl:8023/sdss/dec_result_uris.
The application process takes a few minutes. Execution time depends on the size of the selected fragment and the number of objects contained in the area (of parcels, buildings, networks and roads).
The application selects those individuals (in this case the parcels), which were included in the class which describes expected result of the analysis.
As a result we get the set of parcels, that reasoner classified as type (rdf: type), conforming decision criteria.
For parcels fullfilling such conditions reasoning application updatesa special column in parcels layer table. The PostGIS layer table of the tested objects, contains specially designated for this purpose result column. For each parcel in the result set, in this column, is recorded name of result class.
Styling definition of layers in the desktop GIS project, visualize the result, by coloring of the selected objects.