Hi Bikash,
On Tue, Dec 28, 2010 at 12:39 PM, Bikash Gyawali <bikashg@live.com> wrote:
> I see your point, Marco. So, there's no way to model properties that are
> both locally reflexive and transitive in OWL? In real life, there are
> properties like that, for eg: the hasReferenceTo property or 'Knows'
> property.
>
>
I fail to see why these properties must be transitive. For example If I know
you and you know your father, that doesn't imply that I know your father.
Or, if paperA hasReferenceTo paperB and paperB hasReferenceTo paperC, that
doesn't imply (to me) that paperA hasReferenceTo paperC. At least, it seems
to me that the first two relationships are direct relationships while the
'inferred' relationship is indirect. That is, the correct implication should
be:
paperA hasDirectReferenceTo paperB and paperB hasDirectReferenceTo paperC
implies that
paperA hasIndirectReferenceTo paperC
This way hasDirectReferenceTo can be locally reflexive, but not transitive,
while hasIndirectReference can be transitive but not (locally) reflexive.
You can link the two properties by using property chains. And if you want,
you can also define both direct and indirect properties to be subproperties
of hasReferenceTo. This is a common design pattern in OWL ontologies,
although I couldn't find the pattern in the Ontology Design Pattern portal
[1]. Can someone point to the correct pattern description?.
I've attached a modified version of your toy ontology using this pattern.
Cheers,
Ronald
[1]
http://ontologydesignpatterns.org/wiki/Ontology_Design_Patterns_._org_%28ODP%29