OWL property chains

[Dave Reynolds]

Hi,

Do any of the Jena reasoners support OWL property chains?

No, Jena doesn't support OWL 2 features.

The OWLMicroReasoner seems to support most of what I need except for property chains. I’m wondering if it is possible to extend this reasoner with a property chain rule, or if I should just include the rules I need in a GenericRuleReasoner.

It may be possible.

The route to try is to write a rule set which @includes the owl micro
rules and adds the ones for property chains. You'll need to configure a
GenericRuleReasoner with the appropriate magic settings to run that
extended rule set. See:

https://jena.apache.org/documentation/inference/#RDFSPlusRules

The problem is that the right approach would be to use hybrid rules with
the forward rules triggering off the existence of the property chain
axiom and generating backward rules to implement the axiom. However, I
don't think the forward rules are generic enough to handle the way that
axioms are encoded (with RDF lists). So it might be necessary to do that
with a nasty java hook.

Dave

[Jim Balhoff]

Thank you for the suggestions; it’s helpful to know this isn’t already in there somewhere. So far I’ve implemented simple forward rules for property chains of length 2 and 3, which cover all of the chains found in our ontology, and this is working well. Not general enough for any size chain, but straightforward to write.

Jim

Post by Dave Reynolds
Hi,

Do any of the Jena reasoners support OWL property chains?

No, Jena doesn't support OWL 2 features.

The OWLMicroReasoner seems to support most of what I need except for property chains. I’m wondering if it is possible to extend this reasoner with a property chain rule, or if I should just include the rules I need in a GenericRuleReasoner.

It may be possible.
https://jena.apache.org/documentation/inference/#RDFSPlusRules
The problem is that the right approach would be to use hybrid rules with the forward rules triggering off the existence of the property chain axiom and generating backward rules to implement the axiom. However, I don't think the forward rules are generic enough to handle the way that axioms are encoded (with RDF lists). So it might be necessary to do that with a nasty java hook.
Dave