Re: issue 5.10: a position statement

On Fri, 2002-07-19 at 06:16, Peter F. Patel-Schneider wrote:
[...]
> There are already lots of test cases that have been brought forward to show
> what is needed in the higher levels of the Semantic Web.  On the ontology
> level John belonging to Student and Employee entailing John belongs to
> Employee and Student is a good example,

Yes, in fact that one is much more relevant to issue 5.10.

http://www.w3.org/2002/03owlt/sem-intersectionP.rdf
http://www.w3.org/2002/03owlt/sem-intersectionC.rdf

The sameState/peopleKey tests are reall about issue
 5.1 Uniform treatment of literal/data values
and not right at the heart of 5.10.


After studying the layering issues, my position is
that our semantics shouldn't entail the existence
of any classes from an empty KB (ummm... except
the 'reserved' ones... rdfs:Class, rdfs:Property,
owl:TransitiveProperty, etc.).
So I suggest sem-intersection should be
a non-entailment test.

I stipulate that this is non-intuitive, but
no more so than things like "yes, 
Boeing747 is both a class and an individual,
but the two are not related in any formally
observable way."

We discussed something like this at the ftf...

  ?plane1 rdf:type boeing:B707.
  boeing:B707 boeing:manufacturedSince "1960".
  ==> ?
  ?plane1 rdf:type _:something.
  _:something boeing:manufacturedSince "1960".

which might show up, in practice, ala...

	Show me all the planes in stock
	of some sort that we started making in 1960.



> but there are many others.

Really? I don't recall any others from our
discussions. I'm interested in things that I can
add to the test repository.

>  On the
> logic level A or B entailing B or A is a good example.

I don't understand how to write "A or B" in OWL/DAML+OIL.

Ah... perhaps you mean "the union of the classes A and B",
rather than "the disjunction of the propositions A and B"?
That doesn't seem interestingly different
from the sem-intersection test case, to me; i.e. it
doesn't seem worth another test case. I'm willing to
add it if somebody else says it looks useful.


[...]

> RDF is uninteresting, to me, on its own.
> 
> RDF is only interesting inasmuch as it facilities progress in one of two
> areas:
> 1/ allowing XML data to be used in the Semantic Web, by providing an
> XML-compatible meaning for XML documents that can then be used in the
> Semantic Web;

[more on that later, perhaps in www-rdf-logic...]

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 19 July 2002 11:59:58 UTC