RE: URI etymology

> You can't identify a "namespace" because a namespace isn't a resource
> in the RDF sense of the word - it just is. A namespace in RDF is a
> syntactic thing, not something that you can retrieve over HTTP; just a
> string (that happens to be a URI/URI-Reference to be globally unique,
> and to allow the SW to remain decentralized) 

Absolutely. Yep. Agree with you completely so far.

(though I would say that the namespace is a syntactic thing in the
*XML* serialization, not necessarily in RDF, and has no existence
whatsoever in the RDF knowledge space proper -- i.e triples)

> that when you attach
> other strings to it, comes up with URIs or URI-References that
> identify terms that your processor may or may not understand to some
> certain context. 

Oops. Nope. Can't do that. The logical construct (namespace(name))
does not define any URI reference whatsoever! The fact that the
namespace is identified by a URI reference, and the name potentially
might have meaning as a fragment of some MIME content stream does
in no way that the pair of identifiers map to namespace#name. Nope.
It just isn't so. Sorry.

Even if namespace#name actually is a valid URI reference, is dereferencable,
and the results have anything to do whatsoever with the semantics
ascribed to (namespace(name)), then you are just plain lucky. That
is in no way due to any mechanism inherent in XML, XML Namespaces, or RDF.

Namespace URI refs such as "http://.../foo#" that enable one to get
RDF resource URI refs such as "http://...foo#bar" are hacks. Period.
They work because they are hacks, and the namespaces and names are
carefully crafted to work. But they do not work because that is how
RDF is *supposed* to work or *will* work for any arbitrary namespace
URI reference.

> What we're talking about here is using URLs to identify concepts, not
> URLs to be used as namespaces (although one obviously leads to the
> other). These "concepts", i.e. RDF Resources (or as TimBL calls them,
> "things"), are just things that Semantic Web processors use in certain
> contexts. The context part is the key part that people seem to be
> forgetting here.

Agreed. Namespace URI references are *only* syntactic constructs. (though
because they often are URLs and folks are used to dereferencing URLs it's
really hard to resist ascribing more to them than they really are).

What is also being forgotten or missed entirely is the fact that
syntactic forms as defined by XML and XML Namespaces do not automagically
map 
to RDF resource identities. There is a missing link that is disguised
by the fact that most namespace URIs in use follow the '#' suffix hack 
such that they appear to bridge that gap (though they really don't).

Patrick

--
Patrick Stickler                      Phone:  +358 3 356 0209
Senior Research Scientist             Mobile: +358 50 483 9453
Software Technology Laboratory        Fax:    +358 7180 35409
Nokia Research Center                 Video:  +358 3 356 0209 / 4227
Visiokatu 1, 33720 Tampere, Finland   Email:  patrick.stickler@nokia.com
 

Received on Tuesday, 12 June 2001 08:08:13 UTC