Re: Issue 5.19 Classes-as-instances

On October 18, Jeremy Carroll writes:
> 
> 
> Many months ago Ian asked the very good question what difference does this 
> make?
> 
> Here is an entailment:
> 
> 
> Description: (informative)
> 
> If two URIrefs denote the same thing, then their class extensions are also 
> the same.
> 
> 
> premises
> ========
> 
> first:thing owl:sameInstanceAs first:sameThing .
> 
> conclusions
> ===========
> 
> first:thing owl:sameClassAs first:sameThing .
> 
> 
> ++++++++++++++++++++++++++
> 
> 
> My understanding of the current consensus (being challenged by Enrico) is.
> 
> This entailment holds in Large OWL.
> In Fast OWL and Lite OWL it is (syntactically?) problematic, because it 
> does not satisfy the separation of classes from things condition.

Name separation is not an inherent requirement of Fast OWL, and is not
enforced in the abstract syntax - individuals and classes can have the
same names, but are not logically connected.

Name separation is required, however, if Fast OWL is to be embedded in
RDFS in such a way as to be semantically compatible with Large OWL.

Ian

> 
> Thus, this test case should be added to the test document with the test 
> editors instructed to clarify that this:
> + is a valid large owl entailment
> + should not be expected from a fast owl or owl lite system
> + that a fast owl or owl lite system may indicate that there is an 
> ill-formedness in the question
> + that a fast owl or owl lite system may find this entailment
> 
> And even for a large owl system, one should remember that large owl 
> reasoners will be incomplete!
> 
> I suspect with some care the Test document could support a range of 
> conformance options indicated by metadata tags on the tests ....
> 
> So I am suggesting rephrasing the classes as instances problem into what 
> sort of behaviour do we expect systems to show with this entailment; and 
> how do we document these expectations.
> 
> Jeremy

Received on Friday, 25 October 2002 04:07:16 UTC