Re: status of rdf, rdfs, and owl ``namespace files''

From: Dave Beckett <dave.beckett@bristol.ac.uk>
Subject: Re: status of rdf, rdfs, and owl ``namespace files'' 
Date: Tue, 18 Mar 2003 14:34:33 +0000

[...]

> > What is the status of 
> > 	http://www.w3.org/1999/02/22-rdf-syntax-ns
> > 	http://www.w3.org/2000/01/rdf-schema
> > Are they normative parts of the RDF specifications?  I don't see how, because
> > ...
> 
> I assume you are refering to the documents that you GET at these
> URIs.  Those are handy documents but RDF is a language and neither
> allows nor forbids you to dereference URLs that it uses.
> 
> Brian already confirmed that the document at the second URI was
> normative to you on 3rd January:
>   http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0007.html
> 
> You can see it in the normative appendix of all drafts of the RDF
> Schema/RDF Vocab WD going back several years:
>   http://www.w3.org/TR/rdf-schema/
> 
> With respect to the document at the URI
>   http://www.w3.org/1999/02/22-rdf-syntax-ns
> I agree it does not appear in our drafts, and maybe it should with draft
> updates be made available in a working draft.

OK, so the thinking is that these documents should be normative.  OK, what
then do they mean?  What information needs to be them?

There are several possible answers to these questions, and I do not believe
that there has been much, if any, indication as to which one is supposed to
be the case.

The answers include:

1/ The documents are valid in that they are true in all RDF (RDFS, D-)
   interpretations. 
2/ The documents are complete in that they include all the information
   about the RDF (RDFS, D-) vocabulary that can be represented in RDF
   (RDFS, D-, multiplicatively).  In normal circumstances, this answer
   would generally also include validity, but I suppose that this could be
   relaxed, particularly for rdfs:comment, etc.
3/ They are documents that serve only to introduce all of the RDF (RDFS,
   D-) vocabulary, but no validity or completeness.
4/ The documents only provide some hint as to some aspects of some part of
   RDF. 

As far as I can tell 4/ is the current answer.

> > ...
> > 1/ Neither of them are valid in the RDF Model Theory or the RDFS
> > model theory.
> 
> They are mostly(*) correct RDF/XML documents that generate RDF
> triples but concentrating on validity, what needs to change?
> 
> It seems from the thread above that you mean valid inference from the
> empty RDF graph to one of the RDF graphs produced from the RDF/XML
> here.  The rdfs:comments correction has already been picked up, what
> are the other specific problems.

There is invalid information in these documents related to rdfs:comment,
rdfs:label, rdfs:isDefinedBy, and rdfs:seeAlso.  The RDF file also contains
rdfs:domain and rdfs:range information that is not valid in RDF.  (It is valid in
RDFS, but not in RDF.)

> (*) There may be syntax updating necessary to remove warnings about
> unprefixed attributes - the RDF Vocab WD above contains such a draft
> update.  Plus the new vocab added.
> 
> > 
> > 2/ http://www.w3.org/1999/02/22-rdf-syntax-ns
> >    does not have elements that correspond to all the elements of the RDF
> >    namespace. (read RDF vocabulary here)
> 
> I assume you mean descriptions of vocabulary terms, not XML elements.
> 
> If you trying to say that the triples do not describe the newer
> vocabulary terms that the RDF Core WG added for Collections (List,
> first, rest, nil) and datatypes (XMLLiteral) and are asking for this
> to be corrected, this could be part of the action for the same issue above.

Nope, this is not the only lacks here.

This file does not include any of the container membership properties, for
example.

> > If they are not normative, what is their status?
> 
> The first is already, as in, defined by the current drafts.

As indicated above, normative is not very useful if there is no notion of
what information they need to convey.

> > One reason that I ask is that WebOnt has a similar sort of document
> > 	http://www.w3.org/2002/07/owl
> 
> For some reason the RDF validator fails on that file, might be a
> temporary error.  Seems to generate correct triples when I try it.
> 
> I'm curious if someone has checked if this document is rdf-valid
> and rdfs-valid.

Someone (maybe you) pointed out one problem with the file.  I don't know if
the file has recently been run through an RDF validator.

> > with some of the same problems, but because of owl:imports the actual
> > contents of the document matters more.
> > 
> > In fact, owl:imports makes the contents of
> > 	http://www.w3.org/1999/02/22-rdf-syntax-ns
> > 	http://www.w3.org/2000/01/rdf-schema
> > matter more.
> 
> "matters more" isn't very clear, so I guess this means that OWL has
> some other requirement on these files.  This might need some WG
> coordination if WebOnt the group has some requirement that isn't
> covered by another last call issue.

In RDF and RDFS, there was little formal utility for the two files because
there was very little that could be done with them.

In OWL, it is possible to make the contents of these files be imported into
an OWL ontology.  If these files have problems (invalid, incomplete, etc.)
then these problems can easy be demonstrated in test cases that don't
directly point to the files themselves.  

> So summarising, would these satisfy you for this comment:
> 
>   1. Add a new normative section to one working draft
>      I suggest rdf/xml wd (or possibly vocab) that gives a corrected
>      content of http://www.w3.org/1999/02/22-rdf-syntax-ns
>      with the terms correct for the updated RDF vocabulary terms:
>        first, rest, nil, XMLLiteral
>      probably as a new normative appendix.
> 
>   2. Ensure the two documents at the URIs as given in the drafts
>      are rdf-valid, rdfs-valid from the RDF Semantics WD point of view.
>        - need specific items to track here that have been raised (rdfs:comment)
> 
>   3. Update the documents at the two URIs
>      - this might be a coordination action require some further W3C
>        process, I'm not sure.
> 
>   4. Ask WebOnt what, if any changes, owl:import requires on these documents.

This would be one way to proceed.

Another way to proceed would be to require these documents to have some
other relationship to the RDF/RDFS semantics, perhaps one of the ones
indicated above, and arrange for it to be so.

> Dave

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Wednesday, 19 March 2003 06:59:33 UTC