Re: draft response to #owlref-rdfcore-owl-class-denotation

>Pat "it might be controversial"
>
>Dan had proposed rejecting the comment on the grounds that we did not have a
>worked out design. While I would vote against that; I don't think I can
>seriously degree with the reasoning (other than by working out a design,
>which I haven't the energy for). [A problem is that the lack of an
>alternative design is enough of a motive to postpone an issue, but
>postponing this issue does not make much sense - so it has to be rejected].
>
>Pat's message has lots to disagree with, which can be summed up by
>
>rdfs:subPropertyOf rdfs:subPropertyOf rdfs:subPropertyOf ?

I fail to understand the point being made here. If that is a 
question, the answer is: Yes, so?

>
>I don't find Pat's understanding of OWL DL anything other than mystical. A
>phrase like: "The OWL-DL notion of
>what counts as a class is more limited than the RDFS notion"
>
>no - the two ideas are written down using different marks in different
>formal systems and we cannot compare them

Well, doesnt that strike you as rather unfortunate, given that they 
are both being put forward as notations for use on the semantic web?

>except within the one formal
>framework in which we have made them comparable (OWL Full).

Actually that is simply false. One can use the OWL vocabulary in RDFS 
without accepting the restrictions of OWL Full. People will do this, 
in fact, whether we like it or not; and you know what, when they do, 
it will make perfect sense.

>  In that
>framework they are equivalent (have the same class extension).

OK, that is a serious error in that particular formal framework, IMO. 
The only reason for doing this in OWL Full is to make OWL Full 'seem' 
indistinguishable from OWL DL, and Peter's theorem makes clear.

>
>Any attempt to make such a comparison outside the formal framework is
>mythology: taking some pictures inside our heads a little too seriously.

This may be a disagreement in the philosophy of mathematics. You 
sound like a formalist, Jeremy. I don't think that will be robust 
enough for the SW: like most working mathematicians, I was talking in 
robustly Platonic terms. It seems to me to be obvious that sets 
really do exist which are not in the OWL-DL universe. People talk 
about sets of sets and sets of properties all the time: the OWL specs 
themselves talk about them, every formalized set theory ever written 
allows such things to exist; FOM typically *constructs* models of 
mathematics out of things like this, for goodness sake. In OWL-DL 
such talk is forbidden; but as Galileo might have said: tuttavia, 
esistono.

But less mystically, the point of my claim is that there is a very 
natural way in which OWL-DL can be seen as embedded in RDFS. It is 
the one summed up in a slide I presented at the Boston discussion 
meeting (attached) and which has always been my central intuition of 
OWL-DL as an RDFS semantic extension. The vision of OWL-Full in which 
owl:CLass is identified with rdfs:Class was described there as an 
ALTERNATIVE to the OWL-DL vision; it is in a strong sense 
incompatible with the OWL-DL notion of what owl:Class means. I think 
that what would be more use would be a direct embedding of OWL-DL 
into RDFS which made OWL-DL a semantic extension of RDFS with a 
restricted vocabulary, along the lines of the 'great horned owl' 
proposal I made a while back. OK, I don't expect Webont to actually 
do this at this stage: but all I am asking is that we don't make it 
*impossible* to do this in the future. Making this (IMO) dangerous 
identification of owl:Class with rdfs:Class would block any future 
possibility of making OWL-DL and RDFS fully interoperable and would 
set a huge conceptual mistake in stone.

Pat
-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes