Re: URIs vs URLs?

Jonas Liljegren wrote:
> Well. That was what I was trying to say. That there are a URI for the origin
> of a model and a URI for the model itself.
> I did this separation for clarification. I had always imagined that the model
> URI would be the URI of its origin.
> 
> Since it is important to know if you are refering to a description of something
> or the thing in itself, it could help the understanding to let the URI of the
> description be the description rather than the ting it describes.
> 
> http://abc/jonas could be used as the URI of the person jonas. But this could
> also be the URL of a document describing that person. If that were the case, what
> URI would you use to refer to the document, rather than the person?

I had the same questions about a few tricky situations :

1) About my homepage.
It has author="Champin" and date="12/12/99"
It has both URIs "http://www.cpe.fr/~pa/index.html" and "ftp://www.cpe.fr/home/pa/public_html/index.html"
Though, I could write

  <Decrsiption about="http://www.cpe.fr/~pa/index.html">
    <type resource="http://ontology.org/WebResource"/>
    <s:servedBy> Apache Server </s:servedBy>
  </Description>

  <Description about="ftp://www.cpe.fr/home/pa/public_html/index.html">
    <type resource="http://ontology.org/FTPResource"/>
    <s:owner resource="user://www.cpe.fr/pa"/>
    <s:rights> rw-rw-r-- </s:rights>
  </Description>

And where am I supposed to put author property ? 

QUESTION 1 : does RDF describe the resource entity or the resource content ?
I guess the answer is "both, it just depends", but then on WHAT does it formally depend ?


2) About rdfs:Class and other "virtual" resources.
Many resources are "defined" by ways of RDF.
Their URIs don't return their content (which often doesn't even exist)
but some description of the resource.
That's what Jonas raises :

QUESTION 2 : how do I know when an URI points to the resource, or to a description of the resource
This question raises also when we're dealing with real-world-objet-URIs (like person's),
because those URIs will doubtfully return the resource itself !


3) About services
Differentiating the URL and the URI of the service may be a good thing,
but is it sufficient ?
What about the service "http://www.cpe.fr/cgi-bin/rose"
returning some classical piece of french poetry ?
I can state

  <Description about="http://www.cpe.fr/cgi-bin/rose">
    <author> Champin </author>
    <language> Perl 5 </language>
  </Description>

but the URL will return a documennt with some self-describing RDF inside
    <author> Ronsard </author>
    <form> Sonnet </form>
what URL does this reffer, if not "http://www.cpe.fr/cgi-bin/rose" ??

QUESTION 3 : generally, an URI can have more than one meaning !
(not only container/content or describee/description)
How will we deal with that ?

PS : I'm sure the RDF community could provide me with half a dozen unambiguous URIs,
one for the service, one for Ronsard's sonnet, and so on...
But I'm not so sure all RDF users could agree on all that,
so we WILL have to handle polysemy of URIs...

  Pierre-Antoine

Received on Monday, 20 December 1999 04:17:37 UTC