On 03/12/15 09:52, Dibyanshu Jaiswal wrote:
Hi Dave,
As described in my question, I experimented with a new ontology, and the
results are as expected. Hence, its got be some problem with my original
ontology.
OK, I may not have parsed your question correctly then.
Can you explain (or provide some link) by which you mean "virture of
domain/range inference from their properties." so that i can debug my
ontology. Since I belong to a corporate background, I'm not allowed to
share the ontology. 😑
I mean, for example, if you have some property:
:p rdfs:domain :B; rdfs:range xsd:string .
And then you have an individual:
:a rdfs:type :A;
:p "foo bar" .
Then the reasoner should deduce that :a is also of type :B because the
domain statement on :p means that all things that have a :p property are
within :B.
Similarly for rdfs:range statements.
Dave
On Thu, Dec 3, 2015 at 3:04 PM, Dave Reynolds <[email protected]>
wrote:
On 03/12/15 09:22, Dibyanshu Jaiswal wrote:
Say I have an Ontology (OWL file) with two classes A and B (siblings to
each other, under Thing) with instances A1,A2.. (rdf:type A)and B1,B2..
(rdf:type B) respectively, along with some datatype and Object properties
with A as domain class and B as Range class.
On SPARQL Query
SELECT * WHERE {?ind rdf:type :A}
return me individuals of both A as well as B.. along with the object
properties.
Jena's Inbuilt Infernece Model linke this:
Reasoner reasoner = ReasonerRegistry.getRDFSReasoner();
InfModel infModel = ModelFactory.createInfModel(reasoner, ontModel);
As per my expectations I should get a list of individuals which are of
rdf:type A or rdf:type (subclasses of A)
The chances are those individuals appearing in the SPARQL result which you
believe to be of type B are *also* of type A by virtue of domain/range
inference from their properties.
If you can't spot the problem then isolate a small test case, e.g.
starting with one problem individual and expand from there. If you have a
small test case but still can't see why those inferences are cropping up
then you have something small enough to post to see if others can help you
diagnose.
Dave
