- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Tue, 07 Oct 2003 07:08:12 -0400 (EDT)
- To: seanb@cs.man.ac.uk
- Cc: www-webont-wg@w3.org
From: Sean Bechhofer <seanb@cs.man.ac.uk>
Subject: owl:Ontology triples and DL/Lite
Date: Tue, 7 Oct 2003 11:16:43 +0100 (GMT Daylight Time)
> In the S&AS, an ontology is defined as:
>
> [[
> ontology ::= 'Ontology(' [ ontologyID ] { directive } ')'
> ]]
>
> Additionally, we have the mapping rules that say:
>
> [[
> Ontology(O directive1.. directiven) -> O rdf:type owl:Ontology .
> T(directive1)
T(directiven)
> Ontology(directive1 ... directiven) -> O rdf:type owl:Ontology .
> T(directive1)
T(directiven)
> ]]
>
> This would suggest to me that every DL/Lite ontology represented as
> OWL-RDF must have *at least one* type triple with owl:Ontology as object.
>
> However, many of the test case ontologies don't. For example, picking one
> at random:
>
> http://www.w3.org/2002/03owlt/someValuesFrom/premises001.rdf
>
> has no owl:Ontology triples, but is labelled as Lite (and all our syntax
> checkers agree that it is Lite).
>
> Have I missed something somewhere, or have we all by chance managed to
> successfully not implement the same feature.....?
>
> Cheers,
>
> Sean
As they say, your observations have been overtaken by events.
At a meeting a few weeks ago the working group decided to make the type
triple optional for anonymous ontologies. However, this caused a semantic
layering bug. At its last meeting the working group voted to allow axioms
and facts outside of ontologies in OWL DL and OWL Lite. (Hmm, I have to
check to see if I've written the OWL Lite case correctly.)
peter
PS: The reason that the implementations get it right is partly because it
is just easier to implement a syntax checker that doesn't check that there
is an ontology type triple.
Received on Tuesday, 7 October 2003 07:08:50 UTC