W3C home > Mailing lists > Public > public-owl-wg@w3.org > January 2008

Re: Specifically on ISSUE-65 (Re: Punning, typed vocabulary, and handling RDF graphs (and ISSUE-65))

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Thu, 10 Jan 2008 13:32:26 +0000
Message-ID: <47861E6A.7090605@hpl.hp.com>
To: Bijan Parsia <bparsia@cs.man.ac.uk>
CC: "Web Ontology Language ((OWL)) Working Group WG" <public-owl-wg@w3.org>

Bijan Parsia wrote:
> 
> ISSUE-65 reads as "excessive duplication of vocabulary" and specifically:
> 
> """The member submission documents seem to replace a good many 
> properties from OWL 1.0 with three properties in OWL 1.1. (The old 
> version, and two new versions, one for data properties, and one for 
> object properties)"""
> 
> But now I realize that I don't understand this. In particular I don't 
> understand how this is supposed to be an issue against the RDF 
> serialization. What properties are duplicated? How many is a "good 
> many"? I see some "Classes" duplicated (e.g., ObjectRestriction and 
> DataRestriction), but someValueFrom is someValuesFrom. maxCardinality is 
> maxCardinality.

I think 'properties and classes' would have been a more accurate 
statement of the problem.

Here's a list ...

owl11:FunctionalDataProperty
owl11:FunctionalObjectProperty

owl11:DataRestriction		
owl11:ObjectRestriction

owl11:NegativeDataPropertyAssertion	
owl11:NegativeObjectPropertyAssertion

i.e. 5 classes (the last is new, so one composite class may need to be 
added for the new functionality, cf ISSUE-81)

owl11:equivalentDataProperty	
owl11:equivalentObjectProperty

owl11:subDataPropertyOf		
owl11:subObjectPropertyOf

owl11:dataPropertyDomain	
owl11:objectPropertyDomain

owl11:dataPropertyRange		
owl11:objectPropertyRange

owl11:disjointDataProperties
owl11:disjointObjectProperties

and 10 properties.

This  produces 15 places where there is a three way choice as to which 
one to use[*], and I believe people will find that confusing.

Bijan:
 > In particular I don't
 > understand how this is supposed to be an issue against the RDF
 > serialization.

It is an issue against the design - I think the closest design that 
would not have this issue would be similar to the current design but 
with changes to the mapping rules, and a reduction in punning, I don't 
have a particular position on what should change though. Clearly these 
properties and classes only occur in the RDF, so the mapping rules do 
need to change in order to fix this problem, there may be consequential 
changes elsewhere.

Jeremy

PS. I have copied the list from Jim, who references Peter's
http://lists.w3.org/Archives/Public/public-owl-wg/2007Oct/0190
so some changes may have already been made, or other errors may have 
crept in.

* 14 three way choices, and one two way choice.
Received on Thursday, 10 January 2008 13:32:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 10 January 2008 13:32:56 GMT