RE: Bitzi File Metadata RDF Dump

> A fair point. One component here might be a language for making public
> promises about the quality of service to be expected from 
> certain URIs.
> For example, I would like to be able to say (in a 
> machine-friendly way)
> that I promise to manage certain names at xmlns.com in accord 
> with certain
> policies, that I commit to remembering to pay my DNS fees for all
> eternity, etc etc.

I really didn't want to get back into the thick of this name vs. 
location debate, but...

This "well managed URL" issue is a non-issue for URNs. That's
the *whole* point of URNs. If I define an ontology using URNs, I don't
have to worry about if my server domain name changes, as that
doesn't effect the validity of my identifiers.

HTTP URLs are *supposed* to resolve to some semblence of an online
accessible resource. To use them to identify abstract properties
that either never will have an online realization or whos identity
is independent of, or transcends their online realization, is just 
wrong.

This promotion of HTTP URIs for abstract resources all boils down
to, I think, the desire to have RDF schema ontology namespace
URIs equate to the URLs of actual schema definitions defining
the concepts represented by the names grounded in that namespace.

If I have to pack all possible statements about properties or resources
grounded in a given namespace into a single RDF instance retrievable
from the namespace URI, then no thank you sir. It may work for small
schemas and isolated research projects, but in my experience such
mechanisms just won't scale.

Such a requirement totally flushes the modular many-syndicated-source
power of RDF right down the toilet. Not to mention modular management
of knowledge.

When I see comments such as "hey, your schema namespace URI gave
me a 401 error instead of an RDF instance" all I can think is, 
well, why should you expect otherwise?!

Yeah, I know section C.2 of the RDF spec supposedly says the "schema"
namespace URI should resolve to the "schema" -- but unless what is
really meant is either (a) a default namespace, or (b) an xml:base
defined URI (which actually isn't a namespace ;-) then the requirement
is meaningless. And who says that an RDF instance is equivalent to
a "schema" corresponding to a single namespace. An RDF
instance need not be based on any primary namespace but can
contain statements grounded in an large number of different namespaces
and the sum total definition of a given ontology could reside in an
large number of different RDF instances. 

Expecting namespace URIs to resolve to RDF knowledge for the entire 
semantic web is (at least according to the current specs) a pipe dream.

Long live RDDL... (and when do they start using RDF? ;-)

Cheers,

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 Wednesday, 26 September 2001 14:03:18 UTC