W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 2002

Re: "Resource" (RDF vocabulary definitions)

From: Seth Russell <seth@robustai.net>
Date: Sat, 23 Nov 2002 07:06:38 -0800
Message-ID: <3DDF997E.4030400@robustai.net>
To: Chris Catton <chris.catton@btopenworld.com>
CC: RDF-Interest <www-rdf-interest@w3.org>, "www-rdf-comments@w3.org" <www-rdf-comments@w3.org>

Chris Catton wrote:

>I'm also more than a bit mystified by
>http://www.w3.org/TR/rdf-concepts/#xtocid103660
>
>As I read it, if in my rdf graph I use a fragment http://example.org/#foo
>this points at a foo as defined in the rdf document at http://example.org.
>It is not an option that http://example.org is not an rdf document.
>(the relevant section is 'So when someurl#frag is used in an RDF document,
>someurl is presumed to designate an RDF document. )
>
>I interpret this as having two consequences:
>First, I cannot say something in rdf about a fragment in an html document -
>because it is presumed to be an rdf document.
>
Well, ok, on the one hand A URIREF in a RDF document can identify 
anything whatsoever its author wants it to identify (err as long as you 
own the domain).  All you need to do is to describe it correctly.   
Let's say your not the registed owner of "example.org" and you say in 
one of your RDF documents:

<http://example.org#foo>  rdf:type  <http://example.org#HtmlFragment>.
<http://example.org#foo>  ex:partOf  <http://example.org>.

<http://example.org> rdf:type ex:WebPage.

But on the other hand if you are not the registered owner of 
"http://example.org" and the registered owner comes along and says:

<http://example.org> rdf:type ex:Organization.

... then, the your RDF suddenly contains a catagory error, because it is 
written somewhere that the owner of a namespace gets to define the terms 
in that namespace.   Perhaps the solution is for ExampleOrg to not use 
the same fragments in their web pages as they do in their rdf documents 
and/or for you not to talk about other people's HTML fragments in your 
RDF.  

>Second, I cannot say anything in rdf about a definition in rdf (such as 'the
>comment about the class defined at http://example.org/#foo contains an
>error').
>
Sure you can:

_a:1  rdf:type rdf:Statement.
_a:1  rdf:subject  <http://example.org/#foo>.
_a:1  rdf:predicate rdf:comment.
_a:1  rdf:object "This is a great klass".
_a:1  ex:Contains  ex:SpellingError

>The second consequence may be a pragmatic one.  The first seems very odd and
>probably not what was intended, as suggested by the line ...
>'the RDF interpretation of a fragment identifier allows it to indicate a
>thing that is entirely external to the document, or even to the "shared
>information space" known as the Web'
>which sort of suggests that it might also be possible to indicate a thing
>that is internal to the document - though it's not at all clear how.
>
Well let's face it,  RDF is kind of expanding what webmasters normally 
think of as just an HTML fragment.   Hey, get over it !.

>
>I'd previously assumed that if a fragment pointed at something inside an rdf
>tag then it pointed to the foo as defined in rdf.  
>
Nope, that was covered by "RDF uses an RDF URI Reference, which may 
include a fragment identifier, as a context free identifier for a 
resource."  In other words the URI+FRAG is one long opaque string that 
identifies exactly what is described in the RDF document - nothing more, 
nothing less.  You are not permitted to delve kinto the internal syntax 
of the URIREF.  In RDF,  foo#A does not point to the A part of foo, it 
points to foo#A.  

>If it pointed to
>something inside an html tag, then it pointed at the text (or image or
>whatever).
>So http://example.org/#unicorn pointed at the concept of a unicorn if the
>target of the fragment is inside rdf tags, but at the text that lay within
>the anchor tags if inside an html document.
>
I think the party line here is that the mime type of the document gets 
to define what the fragment means.  Since RDF is supposed to be served 
with application/rdf+xml, then it can define the fragment differently 
than text/html.   One of the problems with this is that a lot of the 
ISP's havent gotten the word to serve documents ending in .rdf  as 
application/rdf+xml ... oh well, the hyper world is not a perfect world.   

>The thread on contexts seems to be floundering around with the same issues,
>and it seems to me we need a much clearer statement of what a URI is
>actually referencing.  Or am I missing something?
>
No I think you have just gouged you finger in the soar.  

PS:  Obviously these views are my own and do not reflect the views of 
the WG of which I am not a member.

Seth Russell
Received on Saturday, 23 November 2002 10:07:28 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:57 GMT