W3C home > Mailing lists > Public > www-rdf-interest@w3.org > June 2004

The semantic web don't need no stinkin namespace names! ;-) (was RE: Mistaken identity?)

From: <Patrick.Stickler@nokia.com>
Date: Thu, 24 Jun 2004 09:09:44 +0300
Message-ID: <A03E60B17132A84F9B4BB5EEDE57957B02A2E9BB@trebe006.europe.nokia.com>
To: <jon@hackcraft.net>, <www-rdf-interest@w3.org>


(prior to leaving on my vacation, I thought I'd toss this one out ;-)


It may be useful to consider that namespace names, while required
to be URIs, are not interpreted as URIs. I.e., namespace names are
syntactic mechanisms, and any meaning associated with a URI used
as a namespace name has no relevance to its use as a namespace name.

I.e., a namespace name URI need not denote a "namespace" (whatever
that is). You can use *any* URI as a namespace name and its use as
a namespace name has *no* affect on or relation to the meaning of 
that URI. Using a URI as a namespace name in no way requires that
URI to denote a "namespace" nor is it incorrect (or even atypical)
for that URI to denote something other than a "namespace".

Thus, IMO, presuming that a URI used as a namespace name will resolve
to any representation having anything to do with a "namespace" is
unfounded by the specs and unrealistic (RDDL included). 

And, as has been pointed out in this thread and countless times before,
there is no reliable means to obtain a namespace name from a URI, so
from the perspective of a SW agent operating on RDF graphs, namespace
names SIMPLY DO NOT EXIST and therefore this is a non-issue insofar
as SW solutions are concerned.

If you want to know about a resource denoted by a URI, query a
knowledge source per that URI. If the resource denoted by that URI
is a vocabulary term and you wish to know about the vocabulary,
then look for statements about the term relating it to the vocabularies
it belongs to, and then query about those vocabularies e.g. to
determine ontologies or content models using such vocabularies, and
subsequently discovering which stylesheets are associated with those 
content models, etc. etc.).

E.g. consider http://sw.nokia.com/FN-1/published

which resolves to a description of a vocabulary term, which includes the
statement

    http://sw.nokia.com/FN-1/published http://sw.nokia.com/VOC-1/partOf http://sw.nokia.com/FN-1/Publication .

telling us that the term is part of http://sw.nokia.com/FN-1/Publication 
which is a vocabulary, and from the description of that vocabulary
can be seen what other terms and vocabularies are related to that 
vocabulary, etc. etc.

Note also that the vocabulary http://sw.nokia.com/FN-1/Publication 
employes terms which do not all share a common URI prefix (what would
equate to a namespace name in XML) because, after all, there
are no namespace names in RDF, only URIs, and what matters is what
terms belong to a vocabulary, not the lexical properties of the URIs
denoting those terms. Keeping the syntactic and semantic layers
cleanly separate allows then for more effective reuse of terms by
functional vocabularies optimized for particular solutions.

C.f. http://sw.nokia.com/VOC-1, a vocabulary for defining vocabularies.

If that vocabulary were used by, e.g. XML content models (DTDs/Schemas),
then those relations could be defined by RDF statements and thus be
discoverable (but in this case, no such relations exist).

All you need are URIs and statements about the denoted resources
using those URIs. And from any term URI it should be possible to
discover what vocabularies, ontologies, etc. the term relates to
and which content models utilize that term, and what stylesheet
or other processing instructions exist for rendering content based
on that term, etc. 

Namespace names are a feature of the RDF/XML serialization and are
syntactic, not semantic, in nature, so let's stop trying to make
them do something they were never intended to do -- and in a way
that violates the principle of URI opacity.

The semantic web don't need no stinkin namespace names! ;-)

Cheers,

Patrick


> -----Original Message-----
> From: www-rdf-interest-request@w3.org
> [mailto:www-rdf-interest-request@w3.org]On Behalf Of ext Jon Hanna
> Sent: 24 June, 2004 06:04
> To: www-rdf-interest@w3.org
> Subject: Re: Mistaken identity?
> 
> 
> 
> > Right, and you have demonstrated that it can be useful to try to 
> > dereference a namespace URI, and sometimes to try to 
> dereference URIs 
> > used as identifiers for rdf resources.  I don't deny that 
> at all, but 
> > it's different from saying that a namespace _denotes_ the 
> dereferenced 
> > document (namespace URIs denote nothing outside themselves)
> 
> I don't claim that a URI (whether a namespace name or not) denotes the
> derefernced document, and strongly disagree with those that 
> do. I do claim
> though that if dereferencing is successful then the document 
> returned should be
> some sort of representation of the thing identified (in this case a
> namespace).
> 
> , or that a 
> > URI used as an rdf resource automatically and always 
> _denotes_ the thing 
> > pointed to.
> 
> An identifier that doesn't identify the identifee is no identifier.
> 
> -- 
> Jon Hanna
> <http://www.hackcraft.net/>
> "...it has been truly said that hackers have even more words for
> equipment failures than Yiddish has for obnoxious people." - 
> jargon.txt
> 
> 
Received on Thursday, 24 June 2004 02:23:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:07 GMT