On 25/04/13 18:27, Joshua TAYLOR wrote:
On Thu, Apr 25, 2013 at 11:46 AM, David Jordan <[email protected]>
wrote:
I am writing a little performance benchmark test to ascertain the
overhead in determining whether a given update is a valid. It was
recommended before on this forum that the best way to do this
efficiently is create a new OntModel A, have it import the
associated ontology, then perform the few updates in this new
OntModel A, and then call A.validate().isValid().
Is an imported model simply a submodel, added by just calling the
add method? I don’t see any specific method for importing a model,
but some documentation suggests that a submodel is an imported
model.
OntModels can have submodels, which are added with
OntModel#addSubModel [1]. However, OWL ontologies can also import
other ontologies via owl:imports, and this is a different concept
than OntModel and submodels.
Not really. Sub-models are what OntModel uses to keep track of imports.
An ontology with imported ontologies is a composite document, sub-models
are the composite pattern analogue of that for OntModels.
If you read an ontology into an OntModel, and that ontology owl:imports
another ontology, the import will end up in a sub-model. Or you can add
a sub-model directly using the Java API. Either way, you end up with the
same composite model structure.
To add an owl:imports to an ontology,
use OntModel#getOntology to get the ontology object for the ontology
that the OntModel represents, and use Ontology#addImport to add the
import. If you go down this road, you'll might need to be aware of
OntDocumentManagers [2] and import processing (whether to load
imported ontologies and the like).
[1]
http://jena.apache.org/documentation/javadoc/jena/com/hp/hpl/jena/ontology/OntModel.html#addSubModel(com.hp.hpl.jena.rdf.model.Model)
[2]
http://jena.apache.org/documentation/javadoc/jena/com/hp/hpl/jena/ontology/OntDocumentManager.html
When the A.validate() is called, it is just going to validate model
A, or will it also validate the submodel, which would include the
potentially large associated ontology?
Validation, and inferencing in general, will work on the whole model. To
demonstrate this, I hacked up a quick gist:
https://gist.github.com/ephemerian/5461867
Before adding the sub-model, the base model is consistent. After adding
a sub-model containing the disjointness axiom, the base model is not
consistent:
Model m0, before isValid() => null
Model m0, before isValid() => true
Model m0, after isValid() => null
Model m0, after isValid() => false
Ian
