Re: exploring ambiguity via the "something-which-has" URI scheme

> >It seems to me that making use of a URI in RDF amount to an owl:import
> >of the URI's associated web content pretty well solves this problem.
> >It would make any incompatible use of a URI looks about the same as
> >self-contradicting RDF graph.
> 
> Yes, but...
> 
> 1.  There isn't any clear notion defined anywhere (AFAIK) of what is 
> 'the ... associated web content' of a URI. (What is the associated 
> Web content of, say, rdfs:Class ?) This is precisely the ball that 
> has been dropped between the SW community and the URI community. Each 
> thinks that it is the other's problem.

bash-2.05b$ lynx --dump 'http://www.w3.org/2000/01/rdf-schema#Class' | head
<rdf:RDF
   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
   xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#">

<rdfs:Class rdf:ID="Resource">
  <rdfs:label xml:lang="en">Resource</rdfs:label>
  <rdfs:label xml:lang="fr">Ressource</rdfs:label>
  <rdfs:comment>The most general class</rdfs:comment>
</rdfs:Class>

......

Can we view this association as like sensing the physical universe? 

Or, on the server side, as affecting the physical universe?

> 2.  owl:import doesn't have any well-defined nontrivial semantics (yet).

What more (less-trivial) do people want than what the OWL docs say?

> 3.  There is no such thing as a self-contradicting RDF graph.

I continue to follow the practice of assuming the semantics of an RDF
graph are the semantics not just of RDF but of RDF in combination with
the semantics of all the terms used in the RDF graph.  Which, IMO, is
the only approach that makes sense in practice.  And brings us back to
my point above.    When I send you a message, the full meaning of my
message depends on the full meaning of all my words; if you don't know
all those words, then you'll miss some of my meaning.   

It's clear to me how that works for web-connected machines if the
semantics of each term are communicated via an axiomatization
available on the web [perhaps using LBase] at the appropriate address;
it's less clear to me how to do this with model theory, or in general
when you can't say what you want in an axiomatization in some standard
language.  But I strongly suspect using a normative rdfs:Comment is
actually sufficient; if you can read the language, you can infer more.

> So while I agree that this picture has a nice intuitive feel to it, a 
> lot of work needs doing before we can make it coherent.

It's good to have work to do.  :-)

     -- sandro

Received on Friday, 9 May 2003 18:51:34 UTC