RE: Summary of the QName to URI Mapping Problem

> -----Original Message-----
> From: ext Stephen Cranefield
> [mailto:SCranefield@infoscience.otago.ac.nz]
> Sent: 16 August, 2001 00:18
> To: www-rdf-interest@w3.org
> Subject: RE: Summary of the QName to URI Mapping Problem
> 
> 
> Dan Connolly wrote:
> > Patrick.Stickler@nokia.com wrote:
> > [...]
> > > B. The RDF QName to URI mapping function is broken and unreliable:
> > 
> > Not so.
> 
> I have to agree with Patrick here.  The mapping works if HTTP URLs
> are used for namespaces  (if you're prepared to live with the
> ambiguity problem that Patrick pointed out, and which Dan is
> not concerned about).  However, it will not work in general for
> other URI schemes.  For example, in a recent post I showed that
> concatenation doesn't work if namespaces are given urn:publicid:
> URIs that represent SGML "formal public identifiers".  This is
> because these must end with a language identifier, e.g. "EN".

It also doesn't work for any URI scheme which has bracketed notation
syntax, e.g. my example urn:partax URI scheme for parenthesised
taxonomies, where 

   xmlns:foo="urn:partax:(foo)"
   xmlns:bar="urn:partax:(foo(bar))"
and
   foo:bar needs to map to the resource urn:partax:(foo(bar)) and 
   bar:bas needs to map to the resource urn:partax:(foo(bar(bas))).

Nobody can rightfully tell me that the above URI scheme is invalid (even
if they wouldn't necessarily choose to use it) or that I must accept the 
invalid and meaningless URI "urn:partax:(foo)bar" (invalid because it
violates the partax URI scheme syntax and meaningless because no such
resource exists) instead of the correct "urn:partax:(foo(bar))" as the 
URI of my resource within any arbitrary RDF context running on any machine
anywhere in the SW. Sorry. Nope. No way.

The greatest challenges facing the SW are (a) genericity, (b) consistency,
(c) non-ambiguity and (d) ontological transparency (equivalence). 

We cannot achieve consistency at the cost of either genericity or
non-ambiguity.
and we cannot achieve ontological transparency without first achieving 
genericity and consistency.

Dispite being consistent, the current QName to URI methodology employed by 
RDF fails to support genericity by discriminating against valid URI schemes 
which may use formats not employing direct suffixation. It fails to maintain

non-ambiguity due to the loss of partitioning between namespace and name.
The standard is therefore broken and must be fixed.

A solution is required which is compatible with *any* conceivable legal URI
scheme, which is required by the standard as the only method used by
compliant
RDF tools, and which cannot potentially introduce ambiguity into the
knowledge
base. 

My proposed rdf:Map element does all that (and more), but I would be quite
happy with *any* solution which meets those criteria. The issue is not about
a "better" solution, or my "favorite" or "preferred" solution -- but about
*any* 
solution that will actually work for the SW in the real world. At present,
there is none.

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 Thursday, 16 August 2001 06:26:11 UTC