DDTF/layering: weak class theory seems good enough (5.3, 5.10)

When Jeremy first proposed[15Mar] avoiding layering
paradoxes by way of a theory of classes
in which "nothing necessarily exists," I replied
that this was going to undercut sharing.

Now I'm not so sure. I was chatting with
some folks, and I tried to explain why
this 'solipsistic' approach wasn't good
enough, and I found that my argument didn't
hold water. It's intuitively appealing,
but I couldn't think of any actual
use cases that need a system that
infers the existence of classes.

On the contrary, I now think I could go a long
way with an OWL language that gave me
inverseOf, UniqueProperty, and the like
without a very powerful theory of classes.

I agree with [15Mar] that the answer
to the Student/Employee test should be:
no, that isn't a valid OWL inference.



[15Mar]
http://lists.w3.org/Archives/Public/www-webont-wg/2002Mar/0179.html

hmm... that doesn't cover all 5 cases from the 21Mar list
of tests...
http://lists.w3.org/Archives/Public/www-webont-wg/2002Mar/0251.html

but that list had some bugs too... there's a debugged
version somewhere, no?

obligatory issues list pointer:
http://www.w3.org/2001/sw/WebOnt/webont-issues.html#5.3-Semantic-Layering

this issue also seems relevant, though it's not yet open:
http://www.w3.org/2001/sw/WebOnt/webont-issues.html#5.10-DAML+OIL-semantics-is-too-weak

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

Received on Wednesday, 29 May 2002 03:16:45 UTC