RE: What to do about namespace derived URI refs... (long)

I think you/we are going about the discussion bassakwards.

(URL | URI | URN) & (MIME type) & (Semantic constraint) are premature.

They are ways to implement desired functionality.  However, that set of
desired functionality has yet to be defined in by participants in this
discussion.  Forget about reaching some sort of consensus on functional
desiderata of unparsed names in RDF and resource pointers.


Forget about implementation issues for now (URL -vs- URI, etc.) and work on
use cases and desired function.

------------------------

Here is what I would want were I an RDF developer.

1) I resent the need to specify a resource type in XML.  In the age of OO
programming I should have the option of passing the application a "void"
pointer and trust (hope?) that the application can resolve the object
reference itself.

(As far as I am concerned names are not things, except when treated as
objects themselves by linguists.  Names are just a species of pointer. [What
about performative utterances? Shhh! go away your bothering me.])

2) As a developer, I will often want to tie meta-data to pointers.

a) I might want to explicitly type the data referenced by the pointer.  XML
used the NOTATION declaration in combination with an ENTITY declaration to
be used in an ATTRIBUTE declaration to do this.  The XML approach should be
adequate for typing in RDF.

b) I might want to tie structural meta-data to a pointer.  I should be able
to do this in XML, and as far as I know I can't.  I regard this as a serious
oversight.  XML is all about structure.  There should be an XML *extension*
expressed in XML to tie structural data to an attribute entity that points
to an internal or external resource.

c) I might want to tie semantic meta-data to a pointer.  This should be
covered by an extension of RDF written in RDF.

d) I might want to tie temporal meta-data to a pointer.  This would indicate
how persistent or ephemeral I expect the referent (and hence the pointer) to
be.  This is neither a problem for the pointer-and-namespace, XML, or RDF
working groups.  Somebody should take responsibility for it though.
	(This goes to the desirability of persistence. I disagree, and think that
we should make no Aristotelian presumptions about the desirability of
persistence, nor should we require conforming RDF documents to make any
commitments to the persistence of a reference or its referent.  However, an
RDF author [and author's of other documents] might want to make an internal
or external qualifications about a pointer or a referent's temporal
properties.
   Thinking about it.  Temporal properties are actually a dimension of
"type" per item 2.a.)

e) I do not want to be straight-jacketed into one naming or pointer
convention.  Maybe I want to make up my own naming convention.  I should be
allowed to do this.
   I should also be able to tell you the *type* of name that I will be
passing to your application (a META-NOTATION attribute if you will).

----------------------

> In my opinion it doesn't matter
> what a 'resource' really is, the only thing that matters is what
> RDF wants to do with it ...

> Regards,
> Stefan

Yes.  But I would go farther and say that following HTML and XML, the RDF
and conforming RDF processors *want* to do nothing with a reference except
pass it to an application that will deal with the reference.
	One thing that the application might do pretty often is attempt to
de-reference the reference and then attempt to retrieve the referent.  It
might also read semantic meta-data qualifying the reference and decline to
de-reference and retrieve the referent. Or it might de-reference the
referent, examine a semantic property of the referent itself and decline to
present the resource.  (Either of the last two cases being ways to censor
access to a site by minors by a user agent.)

Remember, a single utterance (or document, or other resource) is inherently
polysemic.  It does not have *a* semantic schema.  It has a plurality of
potential semantic schemas.  There is also a multiplicity of relationships
between semiologies.

-----------

As a practical matter, I might want to tie a semantic document (or translate
structure with an XSLT) to a third party's utterance (e.g.. structured, laid
out text).  I might want to criticize it, per my rights as a human being.
If the target belongs to an interested agency, they will want to fend off
criticism.  Thus, the semantic analysis will need to be a property of a
pointer, not the utterance itself.  (In fact, regarding the semantic
analysis as a property of the utterance _per se_ is epistemologically
problematic....)

If the utterance can be referred to using a URL is you position that I must
create a URN to point to the URL, then point a URI to the URN?  This seems
perverse puritanism in the design of an artificial language?  To me it seems
reasonable to tie text and RDF information to the URL and let the
application sort it out when passed the data by an RDF processor.


> > > [..] for as long as URLs are around and people can use
> > > them in RDF, they will be used - you can't stop that from
> > > happening, and rightfully so.
> >
> > Well, I'll agree that it's going to be darn hard if not impossible
> > to stop, [...]

> It's going to be impossible. How are you going to stop people? By
> issuing a decree? The only way to get people to stop is if the entire
> world (or a good percentage of it at least) agree that using a URI
> reference who's base just happens to be a URL is wrong for identifying
> arbitrary concepts, and as that is clearly not the case, it isn't
> going to happen.

Received on Saturday, 9 June 2001 01:13:21 UTC