Re: minimal requirements for Arch document

> 
> Dave Reynolds wrote:
> > 
> > Michael Kifer wrote:
> > 
> >> In BLD, if we have a##b and b[foo->bar] then it does not follow that
> >> a[foo->bar]. But a true object-oriented extension (a la F-logic, for 
> >> example)
> >> would add nonmon inheritance and a[foo->bar] would follow.
> > 
> > The possibility that such an extension might be wanted is another good 
> > argument why ## should not be in BLD. Such an interpretation of ## would 
> > be in conflict with other interpretations such as RDFS/OWL and should 
> > use a different predicate.
> > 
> I'm confused now.
> My understanding so far was that the main argument against ##
> was that we do not need yet another subclass relation, since
> there already is one in rdfs and owl.
> But apparently rdfs:subclassOf cannot be extended to cover the
> above case. So apparently we do need something new, which can then
> be extended (in various ways) to cover both,
> rdfs:subclassOf and the above case (and other variants).

I think rdfs:subclassOf or rdfs:type can be extended. But Dave's 
argument is a logical fallacy anyway :-)
I cannot even understand what he means by "should use a different predicate".
## is a different predicate. Otherwise we would not be arguing.


	cheers
	  --michael  

Received on Monday, 22 October 2007 16:13:52 UTC