The Classification GUI is a graphical interface that allows to intuitively assign classes/concepts from an ontology to a set of documents. The intended use is to build up easily and quickly an annotated dataset that may be used for classification experiments. One particularity is the property of multilabel assignments, i.e. it is possible to map a document to several classes instead of exclusively one. The tool was developed within the scope of the ALIS Project.
The following files are contained in the downloadable zip file ClassificationGUI0.02beta_R192.zip.
|ClassificationGUI.jar||The executable with all dependencies already included
|ClassificationGUI0.02beta_R192_src.zip||Source code of the Classification GUI
|res||Directory containing a sample subset of the EUR-Lex dataset. The files are alternatingly html or txt files.
|properties||Properties file defining standard directories etc.
In addition, ClassificationGUI0.02beta_R192_jars.zip contains separately the employed external jars.
Unfortunately, there is no documentation available for the tool. However, it is very intuitive and easy to use:
- Unpack the zip file.
java -cp ClassificationGUI.jar tud.ke.classificationeditor.util.ExtractEurovocOntology res/eurovoc_en.owl
to extract the ontology from the English EUROVOC homepage.
- Launch the tool by typing java -jar ClassificationGUI.jar or simply by clicking on the jar if your operating system is configured appropriately.
- If you extracted the ontology correctly into the res folder, it should now be ready to work like in the screenshots. Otherwise or if you want to use your own dataset:
- Open the source directory (File->Open Source Directory) containing the documents, one (plain or html) text file per document.
- Select the ontology file (File->Set Ontology File) in the OWL format containing the classes.
- Select the class mappings file (File->Set Qrels File) in the qrels format.
SupportThere is no official support, the tool is provided as it is. However, you may contact Eneldo Loza Mencía for any questions.
The tool was mainly developed by Han Dong, a former student worker at our group. Acknowledgments also go to the developers of the used external libraries.