RE: Summary of the QName to URI Mapping Problem

> -----Original Message-----
> From: ext Aaron Swartz [mailto:aswartz@upclink.com]
> Sent: 17 August, 2001 16:03
> To: Stickler Patrick (NRC/Tampere)
> Cc: RDF-Interest
> Subject: Re: Summary of the QName to URI Mapping Problem 
> 
> 
> On Friday, August 17, 2001, at 08:04  AM, 
> Patrick.Stickler@nokia.com wrote:
> 
> > c.f. http://lists.w3.org/Archives/Public/www-rdf-
> > interest/2001Jun/0151.html
> 
> Claim 1: Agreed.
> 
> Claim 2: Agreed.
> 
> Claim 3: You seem to be saying that RDF/XML's use of QNames for 
> properties does not work for all URIs/URI schemes. 

That is what I am saying.

> Agreed, this 
> is on the issue list. This is not an issue with RDF, but merely 
> its XML serialization. 

Agreed. That has always been the issue. But without serialization,
there isn't going to be much interchange of knowledge on the SW,
right?

> Would you be happy with a way of spelling 
> out the property, like the difference between foo:bar and 
> <http://foo/bar> in N3? That seems like the minimum necessary 
> solution. I'd imagine something like:
> 
> <rdf2:propName 
> rdf2:name="http://purl.org/dc/elements/1.1/date">value</rdf2:propName>
> 
> (I'm not suggesting that the RDF Core WG actually add this, 
> merely providing it as an example of a possible solution.)

That would be a step in the right direction, in that one could
bypass the QName issue entirely. But is N3 or NTriples expected
to replace the current RDF serialization model? If not, then
that doesn't solve the problem insofar as global interoperability
is concerned.

> Claim 4: I see no reason why different QNames need to map to 
> disjunct resources. Why is this necessary?

Because QNames are disjunct syntactic names which may have significance
in a much broader scope than just RDF applications and therefore it
is IMO unreasonable for RDF to discard that distinction as something
that need not be preserved.

> However, you seem to go on from these claims to all sorts of 
> crazy redefinitions of RDF.

To be honest, I agree. In hindsight, the proposed mechanisms themselves
were not very good. I was trying to find a way to encapsulate
the idea of explicitly defined mappings between QNames and URIs
but made the mistake of trying to do so with a minimum of new
constructs -- instead suggesting that existing constructs be
reintepreted or overloaded. Clearly, that is not acceptable, and for that
aspect
of my proposal I apologize. Even though the way I tried to capture
what I see as needed (and optional but very useful) functionality,
I hope that rejection of the means does not equate to rejection of
the end (i.e. don't throw the baby out with the dirty bath water).

See my separate posting "A Wishlist of QName::URI Mapping Examples"
for a much more reasonable (hopefully) approach which does not
require any changes whatsoever to either the present RDF spec or
to any existing systems.

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 Friday, 17 August 2001 14:25:13 UTC