Re: third version of semantics document

From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
Subject: Re: third version of semantics document
Date: Thu, 29 Aug 2002 14:34:05 +0200

> 
> 
> Jim:
> > could you make it clear why ...
> 
> Peter:
> > The first paragraph of section 1 of the document states that individuals
> > are not classes or properties.  QED
> 
> While Peter's answer may be logically correct I think it demonstrates a
> stylistic tension between a document driven versus an issue driven process.
> 
> I get the impression that Peter sees a process in which a document (e.g. his
> semantics doc) specifies a coherent view, and then the issues are resolved
> by deduction from that document.
> 
> An alternative process (at the other extreme) is that we have a range of
> issues, we make informed (but piecemeal) choices about those issues and then
> try and create a coherent document that encapsulates those choices (an
> abductive process). Obviously coherency may be hard or impossible, in which
> case the conflicting choices need to be revisited.
> 
> A third process, which is what I thought we were following, is that we have
> some provisional documents that give a more or less coherent view; and we
> use those to help inform piecemeal choices about issues. Where we choose to
> not follow the consequences of those provisional documents, then changes to
> the documents will be necessary.

If only it were so.  At least one provisional document [1] on many
of these issues was submitted to the working group over five months ago.
Two very different provisional documents [2] [3] on these issues were
submitted well over a month ago.  No choices have resulted.

> Thus, I would find an answer to the rdf:Class versus owl:Class question as
> one that gave a clearer indication of what problems we solve by the
> additional complexity of having two distinct Class concepts.

To answer this in general requires perusal of the semantics.  

My semantics does not mention owl:Class at all.  The only place that
owl:Class shows up in my documents is in the translation from the abstract
syntax to triples.  Pat's semantics uses owl:Class as the domain and range
of various OWL properties, as well as for other things.  His semantics
could be rewritten to eliminate owl:Class, at some loss in clarity, I
think.


> Jeremy


[1] Semantics for the Proposed OWL Knowledge Base Language
http://lists.w3.org/Archives/Public/www-webont-wg/2002Mar/att-0238/01-semantics.text

[2] Model-Theoretic Semantics for OWL
http://lists.w3.org/Archives/Public/www-webont-wg/2002Jun/att-0081/01-semantics.html

[3] a first-order same-syntax model theory
http://lists.w3.org/Archives/Public/www-webont-wg/2002Jun/0152.html

Received on Thursday, 29 August 2002 08:59:53 UTC