The tool, which can be downloaded here, is used to generate complex ontology mappings between two ontologies.
It has been created within my Bachelor Thesis .


ComplexMappingWindows.zip ( 3,46 MB - MD5: c482511c33bc02cdd7c66eb9bc368872)
ComplexMappingLinux.zip ( 3,46 MB - MD5: c34d265d94922c135ea1ab62b36c968e )
Javadoc.zip ( 330 KB- MD5: 7464f7ceb0b21afc50aad6524c8a0b13 )
SourceCode.zip ( 6,01 MB - MD5: 424632b21b088c1280a08a88f0cfb40c )
Thesis.pdf ( 1,33 MB - MD5: a97e4028ebaba4de1db13826f8aa4fc3 )

Download the corresponding zip-file for Windows or Linux and extract it. The folder contains a script file to start the application via the command line and a jar-file. Other files in the folder are necessary for computing complex mappings, but not appreciable for the user. An installed java virtual machine and a correctly set environment variable is required.

To start the application, open a shell and switch to the directory where the folder, containing the starting script, is located. Execute the script file whereby three or four parameters have to be passed.
Windows:
D:\ComplexMappingWindows>run.bat ontology1 ontology2 referenceAlignment [fileToWrite]
Linux:
x@y:~/D/ComplexMappingLinux>./run.sh ontology1 ontology2 referenceAlignment [fileToWrite]

The first two parameters ontology1 and ontology2 are the paths to the ontologies of these the complex mapping should be created, the third one referenceAlignment the path to the reference alignment. All three parameters are necessary and cannot be omitted. To get a file containing the complex correspondences and not only to see them on command line, fileToWrite is optional and specify the path to a file where the correspondences should be written to. Only the URIs in shortened form are outputted anyway for a better readability.

An example for Windows:
D:\ComplexMappingWindows>run.bat cmt.owl ekaw.owl cmt-ekaw.rdf Entering this command, the complex correspondences between the ontologies cmt and ekaw will be computed and printed, if files cmt.owl, ekaw.owl and cmt-ekaw.rdf are located in the same folder as run.bat.

Computing the complex alignment, especially the similarity between names, is based on the following thresholds: 0.7 for Qualified Restriction and 0.6 for the other types.
This seemed to be the best parameters received through the tests. Additionally a 1:1-Mapping is implemented whenever it is possible. Sometimes it is not realizable because it decideable which correspondence is the better one. In that case a strictly 1:1-Mapping could not be ensured.

If you have questions or like to have further information, please contact me:
Dominique Ritze
dominique.ritze@students.uni-mannheim.de