semantic layering document for the F2F

In trying to advance towards compromise and consensus, I seem to have 
walked into a hornets nest.  I now think that the draft I was 
preparing was more contentious and less of a compromise than the 
previous draft. I therefore suggest that we revert to the previous 
draft as the 'locked down' version for discussion at the F2F, that is:

http://www.coginst.uwf.edu/users/phayes/RDFS2OWL-L.html

HOWEVER, please bear the following in mind.

(1) the weak/large/fast OWL names were never intended for use in a 
reference document for publication, only for our own internal 
discussions. Particularly 'weak' (so called only because one gets it 
by 'weakening' the semantic closure conditions.) I will go on using 
the terminology here, however, just to save more writing, OK?.

(2) Please try to keep editing and presentation issues separate from 
technical issues. I know this isnt the kind of write-up we would want 
to present to the outside world, being full of alternatives, in-jokes 
and so forth. Also there may well still be tweaking bugs in the 
details of this document. I suggest that anyone who finds them just 
sends them to me and I will fix them without any fuss. Peter has 
already found a few nits which are not in this draft, of course, but 
I will not alter it now until the F2F.

On that point, the semantic conditions in that document are presented 
in a messy way, scattered around the document. The newer version 
(below) shows a neater way to do it, and Im sure we can improve on 
that some more.

(3) When I was writing it, it was meant to present large and fast as 
the two alternatives, rather than weak and fast; 'weak' is just a 
name for what they had in common. Since then, the discussion has 
centered around the idea of having fast and weak as the alternatives. 
This seems to me to be pointless.

I can see the rationale for sticking to the fast option, given its 
computational advantages and so forth. I don't like the idea much, 
but it does make sense. But trying to do fast and weak doesnt make 
any (technical) sense at all.  There really is no technical rationale 
for offering the weak option as an alternative form to the world at 
large. From the users point of view, the large option is simpler: it 
requires less care to use, uses fewer triples, has a smaller 
vocabulary and provides more useful entailments. It is easier to 
explain, and it merges seamlessly with RDFS. It also conforms more 
cleanly to the RDFS world-view, if I can be permitted to invoke such 
a fuzzy criterion. Of course the large version doesnt fit into the 
OWL abstract syntax, but then neither does the weak one, for much the 
same reasons. If you want to invoke an engine built by Ian or Peter, 
you have got to tighten your corsets and put on your armor. The only 
- I emphasize, ONLY - argument for the fast +  weak and against the 
large + fast combination of options is that Peter Patel-Schneider 
doesn't believe that the large semantics is mathematically coherent. 
Under normal conditions this is a fact I would be happy to simply 
ignore, but given that we are all in one WG I will undertake to give 
a mathematical proof that it is coherent. But I do not think that it 
makes sense for us to tell the world that everyone has to keep 
writing 'foo rdf:type owl:Class' on the grounds that Peter isn't 
comfortable with NWF set theory.

I will continue working on the other document at the same uri:

http://www.coginst.uwf.edu/users/phayes/RDFS2OWL-F2F.html

but I probably won't get it done before the end of the weekend, as I 
have a lot to do before about a month travelling. What is (will be) 
there, together with the earlier emails I sent, will probably give an 
idea of the style and overall structure I was going to adopt, but now 
think was probably a bad idea.

Pat

-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola               			(850)202 4440   fax
FL 32501            				(850)291 0667    cell
phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes

Received on Friday, 27 September 2002 21:56:46 UTC