W3C home > Mailing lists > Public > www-tag@w3.org > March 2002

Re: The range of the HTTP dereference function

From: Tim Berners-Lee <timbl@w3.org>
Date: Wed, 27 Mar 2002 16:27:44 -0500
Message-ID: <0aef01c1d5d6$3c35c900$0301a8c0@w3.org>
To: "Jonathan Borden" <jonathan@openhealth.org>, "Roy T. Fielding" <fielding@apache.org>
Cc: <danc@w3.org>, "'www-tag'" <www-tag@w3.org>
> > > [Tim:] In this way, Resource in URI and Resource in RDF can be
similarly
> anything,
> > > but we have an important concept of a <part of the Web information
> space>
> > > <document?> or whatever.
> >
> > [Roy?] Which is false.  RDF is broken if it cannot describe the Web in
its
> entirety.

False that the concept of a document is important?
RDF can describe anything without HTTP URIs.

> I am not at all sure that the idea that "http://..." URI need identify a
> document is at all an RDF concept.

It isn't.  It follows from the the HTTP spec.  Like an ftp: URI identifies
a file.

> Rather I ask Tim how one could say _in RDF_ that "http:" scheme URIs
> necessarily identify documents,  in a way that an interencing engine that
> understands _only_ RDF could understand?.

(Short answer; you can't, it is a level breaking concept.  It is like
telling a browser not to display web pages where TCP protocol
has had to reqest more than three retransmissions.)

Well, I'd have to understand what you mean by "understands _only_ RDF".
Taking this strictly, the inference engine would have onlythe axioms
of RDF, not of RDFS or DAML or anything else, and would understand
no terms not in the RDF spec.

Taking it more liberally, that to "understand RDF" means to understand the
meanings of the properties which you come across, we could
I suppose make a special level-braking Property
"class of things to which everything with the following as a URI prefix must
belong"
(cteupb) and declare

contact:Mailbox foo:cteupb  "mailto:".
http:Document foo:cteupb "http:".

or words to that effect.  I'm not sure why we are doing this excercise.
The cwm level-breakers can be used to define foo:cteupb it in a rule,
but that isn't just RDF.

> i.e. one is not allowed to
> introduce a function "starts-with(URIref)" that RDF is expected to
> understand... since RDFCore removed rdf:aboutEachPrefix, there is no such
> construct, at least none I am aware of.
>
> The bottom line is that neither the RDF or DAML (hopefully soon to be OWL)
> model theories contain operators that allow one to make assertions, or
> derive inferences, from parts of URI references. URI references are
treated
> as opaque names.

That is how it should be.  I wouldn't want those systems to have
such messy leaks.

> > The whole point in changing the model of the Web from a collection of
> > Universal Document Identifiers that can be used to retrieve documents to
> > one where Uniform Resource Identifiers are used to identify resources
that
> > can be accessed in the form of a representation through a uniform
> interface
> > was so that we could accurately model how people have been actively
using
> > the Web since computational hypertext was introduced in 1993.

Excuse me, computational hypertext, if you mean what we called "virtual
hypertext"
was introduced in the form of the CERN phone book in in 1991 before the web
software was announced on usenet.  What was introduced in 1993?

The "documents" have always been abstract things. But abstract information
objects.
What "change" actually occured?
How can you change soemthing to model the way it has alwways been?


I think you are taking my meaning of the word "document" too strictly.
I would include that called "The Bible" as a document.  I would
not call my car a document.  A picture of it, yes.
Its home page, yes. Not the car.

tim
Received on Wednesday, 27 March 2002 16:26:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:05 GMT