Re: rule rdfD4 not correct, may affect test case

I agree with Pat and hope Jan can fix this before publication.

Jeremy

pat hayes wrote:

> 
> Jan, my apologies for not grasping the point of your message earlier.
> 
> The rule rdfD4 (when ddd's value class is a subclass of eee's value 
> space, you can conclude ddd rdfs:subClassOf eee ) isn't actually valid 
> any more, since we made subClassOf non-extensional.  So I think we 
> should delete this rule.  This will affect the xsd: test case we were 
> tal.king about recently, since now it is *never* valid to conclude a 
> subClassOf relation from the empty graph.
> 
> An alternative would be to modify the semantics of datatyping to insist 
> that they are treated extensionally, but that would be hokey.
> 
> Another cheap fix to the test cases doc would be to say in that case 
> that we were assuming the extensional interpretation for subClassOf, but 
> that is kind of tacky since the semantics doc says explicitly that those 
> assumptions are not rdf/s/D-valid.
> 
> Sorry I didn't catch this earlier, guys, particularly as Jan already 
> caught it and I told him to drop it.
> 
> Pat

Received on Tuesday, 2 September 2003 04:36:06 UTC