W3C home > Mailing lists > Public > www-webont-wg@w3.org > February 2003

Re: annotations was Re: MINUES: Teleconference 30 Jan 2003

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Sat, 1 Feb 2003 21:09:11 +0100
To: www-webont-wg@w3.org
Message-Id: <200302012109.12026.jjc@hpl.hp.com>

>> ... I have already proposed a small inelegant fix [1], which you have sneered 
>> at. I would value having my proposal demolished.

>I don't see a solution there.  For example, how does this allow for
>annotations on classes or properties in the direct semantics?

It does not, and does not need to.
The annotations on classes or properties are ignored in the direct 
semantics. 

Our job does not concern the abstract syntax and semantics, but the 
exchange syntax and its semantics.
It is a convenient tool to use the abstract syntax and its semantics
because the abstract syntax conforms better with the syntax the DL
community uses, and the direct semantics borrows from work that
has been done in that field.
It seems clear that that work does not cover annotations well.
Annotations are the bread and butter of RDF, and so it should be
no surprise that the semantics of RDFS works fine on annotations.

The current position in AS&S is that OWL documents within
the OWL DL subset are given their semantics using the direct
semantics on the abstract syntax form; and this does not work
for annotations.

The meaning of documents can be seen in terms of interpretations,
and in this sense OWL Full shows how to give a meaning to an OWL
DL ontology with its annotations.

An alternative view, would be to see the meaning solely in terms
of consistency and entailment.
With appropriate syntactic constraints, OWL DL annotations cannot
cause inconsistency; so the direct semantics is fine for the definition
of consistency. This alternative view is seen in the correspondence
theorem.

My paragraph shows how to combine the entailment of the direct semantics
with the entailment of RDFS (for annotations only).

>> Also, the correspondence proof needs to be overhauled. (see [2]), so that is 
>> no new work.
>
>The flaws in the proof have already been fixed.

But the flaws in the theorem (not covering annotations and imports) [2] seem
fundamental!

Jeremy  

[1] section headed: "OWL Lite/DL Entailment" in
 http://lists.w3.org/Archives/Public/www-webont-wg/2003Jan/0523.html

[2] Herman's AS & S Review: Appendix A.1:
http://lists.w3.org/Archives/Public/www-webont-wg/2003Jan/0424.html
Received on Saturday, 1 February 2003 15:08:17 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:57 GMT