Hi Bonnie I will try to draw the complete picture in this mail.
It is perfectly true, that you can run KIM with a completely new ontology. Importing the RDF in OWLIM is only the first part of the process. After this you will have to make the resources you use for IE aware of the new ontology. For example changing the way the Gazetteer fills its dictionary, making grammars aware of the new classes (Lookup.class feature) etc. That way you will be able to extract the new entities. Extraction is one of the modules of KIM. Over that, we provide indexing and search mechanisms. Searches are accessible though the web interface and on low level though the Java API. But because the WEB UI is tightly related PROTON, you will not be able to use it. You will have to use the Java API to search for entities and documents. So using a completely new ontology is a lot of work and efforts. KIM makes it easier if you map your ontology to PROTON. That way you can reuse the already existing mechanisms and save a lot of work. Reading this resource carefully will answer most of the questions how to do that: * http://ontotext.com/kim/doc/KimDocs-3.0-EN/ExtendInformationExtraction.html The minimum requirements are: - your classes should inherit protons:Entity directly or indirectly - make your classes and properties visible by setting kimso:visibilityLevel1 - provide at least a MainAlias* to all the instances * you can use either aliases or labels to describe your entities. If you use labels, the property is protons:mainLabel . Which mechanism you use is described in KIM/config/install.properties (com.ontotext.kim.KIMConstants.ENTITY_DESCR property). It is generally a very good idea to subclass the main classes in PROTON * protont:Person * protont:Organization * protont:Location This will make them visible for some of the resources. Another thing to have in mind, is the default behavior of the gazetteer. Only entities generated by trusted sources will be included in the dictionary. That is, the below RDF is valid for them: ?entity protons:generatedBy ?trusted . ?trusted rdf:type protons:Trusted . You can add your own trusted source or use one of the already existing. You can also provide a custom query to fill the dictionary of the gazetteer. This is done by putting a text file with the query (sparql or serql) in KIM/config/query.txt . When using some ontology modeling tool (like Protege), it is enough to include the proton modules and follow the rules above to allow KIM to make a good use of the ontology. And again, all the information you need and even much more detailed is in the documentation: * http://ontotext.com/kim/doc/KimDocs-3.0-EN/ExtendInformationExtraction.html Hope this helps, Philip Alexiev Software Engineer, KIM Team _______________________________________________ Kim-discussion mailing list Kim-discussion@ontotext.com http://ontotext.com/mailman/listinfo/kim-discussion
