W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > May 2002

Re: RDF Graph questions

From: Brian McBride <bwm@hplb.hpl.hp.com>
Date: Wed, 29 May 2002 16:46:04 +0100
Message-Id: <5.1.0.14.0.20020529150847.02cc8a50@15.144.25.13>
To: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>, rdf Core <w3c-rdfcore-wg@w3.org>

Sans Chapeau

At 10:39 24/05/2002 +0100, you wrote:
[...]

>2. Can any URI ref be a property name or must there be some associated
>namespace?

I'm concerned that this question is not quite capturing the issue.
As DaveB has pointed out, any URI ref can be associated with a namespace
using rdfs:isDefinedBy.

I think there is consensus that there is no syntactic restriction on the
form of URI that can denote a property in the graph (though some graphs
may not be serializable in RDF/XML without some schema based trickery).

M&S does, in my view, clearly say that a property must be "associated" with
a namespace.  In answer to Pat's question, my interpretation of that, is that
M&S states it should be possible to determine what namespace is associated
with a property.

We may choose to clarify those statements in M&S.

I propose the following test case.

Are the following two graphs, equal?

   <rdf:RDF xmlns:foo="http://example.org/b">
     <rdf:Description>
       <foo:ar>foobar</foo:ar>
     <rdf:Description>
   </rdf:RDF>

and

   <rdf:RDF xmlns:foo="http://example.org/ba">
     <rdf:Description>
       <foo:r>foobar</foo:r>
     <rdf:Description>
   </rdf:RDF>

Our present specs would say that they are.

My reading of M&S is that it says that these two fragments of RDF/XML do 
not represent the same information as they 'associate' the uri 
http://example.org/bar with different namespaces.

Thus a "clarification" is clearly in order.

It seems that there a number of possible positions, including:

   1) at least one of the above fragments of xml must be wrong, as the
      same uri can't be 'associated' with two different namespaces

   2) the above fragments of xml are "equal".  (A different mechanism must be
      used to determine the namespace associated with a URI.)

Option 1 is closer to (my reading of) M&S.  I think option 2 commands more
support in the WG and is I think, better.

Propose:

  o The two graphs described in the above test case are equal

  o Resources which are RDF schema properties and classes MAY be described
    in one or more associated schemas.

    Designers of RDF schemas SHOULD identify their schemas with a URI
    reference which ends in a character which cannot be part of an XML
    localname, such as "#" or "/".

    Designers of RDF schemas SHOULD identify the resources defined by
    a schema with a URI consisting of the concatentation of the URI REF
    of the schema with an XML localname.

    The resource identified by applying algorithm A (described below) to
    URI U MAY contain RDF schema information about the resource identified
    by U.

    The value of an isDefinedBy property of a resource MAY contain RDF
    schema information about that resource.

Algorithm A is the usual namespace splitting algorithm.

The above need wordsmithing/refining, if the approach find favour with the WG.

Brian
Received on Wednesday, 29 May 2002 11:46:58 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:48:16 EDT