- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Wed, 26 Mar 2008 15:50:11 +0000
- To: Michael Smith <msmith@clarkparsia.com>
- Cc: Alan Ruttenberg <alanruttenberg@gmail.com>, public-owl-wg@w3.org
On 26 Mar 2008, at 13:45, Michael Smith wrote:
> On Sun, 2008-03-16 at 21:30 -0400, Alan Ruttenberg wrote:
>> RDF/XML serialization for anonymous inverse properties
>
>> I think it is beneficial to allow for anonymous inverses in
>> ObjectPropertyAssertion.
>
> Can you elaborate this? If we don't have a real requirement for this,
> the simplest resolution* to ISSUE-86 is to require named object
> properties in object property assertions.
I thought of this solution and was unhappy.
> I.e., change
>
> objectPropertyAssertion := 'ObjectPropertyAssertion'
> '(' { annotation } objectPropertyExpression sourceIndividualURI
> targetIndividualURI ')'
>
> to
>
> objectPropertyAssertion := 'ObjectPropertyAssertion'
> '(' { annotation } objectPropertyURI sourceIndividualURI
> targetIndividualURI ')'
This seems unfortunate since I may then be required to introduce
names for inverse properties *solely* due to abox statement. Indeed,
if I *anticipate* people *ever* having to muck about with inverse
property assertions, I'm forced to name *all* my inverse properties.
It also means that I can't have collections of abox statements that
transparently mimic the class expression structure, i.e., I might have:
a:some.p(some.p- C)
(I.e., a has a p successor which has a p precessor which is a C)
but not:
a p _:x. _:x p- C.
or
a p b. b p- C.
Instead, I'd have to introduce an axiom naming p-. I think that's
*much* less usable, not just because of name bloat, but because when
you are describing relational structures it's often much easier to
conceptualize it "navigationally". In general, inverse property
expressions make the language more compositional, which, obviously, I
like.
It is a place where the RDF is tricky, at best.
Cheers,
Bijan.
Received on Wednesday, 26 March 2008 15:48:40 UTC