Re: URIs for terms: motivation [was: Requirements Document]

From: Pat Hayes <phayes@ai.uwf.edu>
Subject: Re: URIs for terms: motivation [was: Requirements Document]
Date: Fri, 15 Feb 2002 14:58:53 -0500

[...]

> Ok, let me stake out another middlin' area that I think we do need to 
> at least think seriously about. We use URIs in at least two distinct 
> ways. They are used as logical constant symbols; and they are used to 
> identify documents which contain ontological content written in some 
> language. Right now we just smush these together; I'd like us to at 
> least think about whether we ought to keep these uses separate, if 
> only conceptually, and what the relationships between them are.
> 
> Pat

This is roughly what I was trying to get at with my comments about
ontologies as resources.

In particular, I would be very worried if OWL could do things like

<In document http://x/o.ont> 

http://x/o.ont rdf:type owl:ontology .
http://x/o.ont rdf:type foo .
foo rdf:subclassOf bar .
bar type owl:Restriction .
bar owl:onProperty owl:imports .
bar owl:toClass bbb .
bbb owl:oneOf :_1 .
:_1 owl:first http://x/a.ont .
:_1 owl:rest :_2 .
:_2 owl:first http://x/b.ont .
:_2 owl:rest :_owl:nil .
foo rdf:subclassOf baz .
baz type owl:Restriction .
baz owl:onProperty owl:imports .
baz owl:minCardinality "1" .

http://x/a.ont#John rdf:type http://x/a.ont#B .

and have this have the meaning that http://x/o.ont imports either
http://x/a.ont or http://x/b.ont.

Worse, what if http://x/a.ont contains

http://x/a.ont#A owl:disjoint http://x/a.ont#B .
http://x/a.ont#John rdf:type http://x/a.ont#A .


peter

Received on Friday, 15 February 2002 15:17:36 UTC