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

Re: Agenda for TC 2008-06-04

From: Ivan Herman <ivan@w3.org>
Date: Wed, 04 Jun 2008 16:02:27 +0200
Message-ID: <4846A073.2020203@w3.org>
To: Alan Ruttenberg <alanruttenberg@gmail.com>
CC: OWL Working Group WG <public-owl-wg@w3.org>


Alan Ruttenberg wrote:
 >        Issue 109 XML Schema namespace - new or reuse RDF namespace. 
We seem to be at an impasse.

(I think you meant OWL namespace)

I am not sure it is as bad as that, except that Bijan and I have 
diverged a bit to other issues, too (see thread starting at [1]). I 
opened the door for this, for which I apologize. On the other hand, our 
discussion also helped in clarifying some points (at least in my mind). 
Let me try to summarize the status for this evening.

I believe the situation is as follows. We have two different tasks for 
which we need an agreed upon URI:

1. as an XML Namespace URI for the OWL/XML serialization
2. as a prefix for the URI-s to be used in defining the vocabulary of 
OWL in RDF (eg, blabla#sameAs, blabla#Thing, etc).

I think we've already agreed that the URI to be used for #2 is

http://www.w3.org/2002/07/owl#

ie, this is the prefix value to be used when encoding OWL in Turtle, 
RDF/XML, etc, as well as defining the semantics of OWL Full and OWL-R.

The issue is whether the same URI should be used in the #1 sense. It is 
true (and my uncertainties are gone) that, formally, this URI _can_ be 
used as an XML Namespace, too (ie, according to the Namespace 
specification), in spite of the trailing '#'. It is, however, important 
to realize that the two usages of the URI have a very different roles.

 From that point on the disagreement between Bijan and me is, I believe, 
a kind of a judgement call:

- Bijan believes that introducing a _different_ URI for the purpose of 
#2 is too expensive, so to say, in terms of the user community, and that 
issue of this extra 'price' should have a higher priority than other 
considerations

- I am concerned that mixing two very different features/roles on the 
same URI is not a clear design and may be misleading (see also my remark 
below), and I do not feel the 'price' referred to by Bijan to be high 
enough to overrule this concern.

I hope Bijan agrees that this is what our discussion boils down to. If 
so, I think the WG can make its choice.

If you still have the patience of reading:-) some more remarks:

1. The thread in [1] slightly diverged towards the question on what 
should one find when dereferencing the namespace and prefix URI(s). This 
is _not_ part of this issue and we should put this discussion aside (and 
probably come back to it at a later stage)

2. It must be said that Bijan's approach already has a precedence. 
Indeed, the URI

http://www.w3.org/1999/02/22-rdf-syntax-ns#

is used in two different roles:

- used as a prefix for the definition of the RDF vocabulary like 
rdf:type, rdf:Property[1]
- used as an XML Namespace for the definition of RDF/XML attributes like 
rdf:about or rdf:resource or indeed rdf:RDF

Ie, if we accept Bijan's proposal, we would follow the decision taken by 
the RDF group years ago for the case of OWL.

Except that... in my opinion that decision was wrong and led to extra 
problems. It was (partially) responsible for the general mixup between 
the RDF Model on the one hand and its serialization in XML on the other 
which, in my view, created lots of problems in practice and in the 
acceptance of the technology. The most blatant example is the rdf:li 
term which is _not_ an RDF vocabulary element but in the XML namespace; 
lots of people fell into the trap of expecting a triple with an rdf:li 
property when querying a triple store in, say, Jena or RDFLib. Won't we 
run into the same problem in having to explain the RDF community that 
owl:Declaration is _not_ an OWL vocabulary term, although they might see 
it in some of our documents, and that it should _not_ be used in the RDF 
encoding of an OWL ontology, because owl:Declaration is merely used in 
the OWL/XML serialization?

I know the analogy is not 100% o.k., because the role and usage of 
OWL/XML is different from RDF/XML; and that most of the RDF users would 
not even see the owl:Declaration (XML) term in their daily work. But it 
is still a reason of concern. Again, this is my own opinion, not 
necessarily Bijan's:-)

3. If we decide to have a separate XML Namespace URI, I also proposed

http://www.w3.org/2002/07/owl/xml

as an alternative which, though a bit long, makes the relationship 
between the two much clearer.

Sorry for the longish mail (again...)

Ivan

Tracker: this is related to issue-109

[1] http://lists.w3.org/Archives/Public/public-owl-wg/2008Jun/0017.html
[2] http://www.w3.org/TR/rdf-mt/#RDFINTERP
-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf
-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf


Received on Wednesday, 4 June 2008 14:03:01 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 4 June 2008 14:03:01 GMT