W3C home > Mailing lists > Public > www-rdf-interest@w3.org > May 2004

Re: Placing a comment on a relationship?!

From: Graham Klyne <GK@ninebynine.org>
Date: Thu, 20 May 2004 11:22:52 +0100
Message-Id: <>
To: "Richard Lennox" <listserve@richardlennox.net>, <www-rdf-interest@w3.org>

First, a question.  Do you mean a comment on the property in general, or on 
the particular instance of the property that relates Class1 and Class2?

For what follows, I'll assume the latter, because the first is a simple 
"yes, of course".

Three approaches come to mind:

(1) reification.  Reification has caused a good deal of confusion, and is 
somewhat clunky, but it's the only option that's currently part of the 
formal RDF specification.  Reification has meant different things to 
different people over the years, and in formalizing the reification 
semantics (which are quite weak) the RDF WG had to make a choice, and came 
down on the side of a form of reification that could support provenance 
information (and I regard your comments as a weak form of provenance).  I 
can't remember the semantic details, but the relevant (non-)entailments are 
mentioned in the RDF formal semantics specification [1] (section 3.3.1)

(2) There is some current research interest in the idea of named graphs 
(e.g. [2]), which also seem to appear, in different forms, in many 
different software implementations of RDF.  Rather than applying comments 
to an individual RDF statement, as (1) above, they would be applied to a 
graph (denoted by an RDF node), with the possibility of additional 
syntactic structures in RDF to associate a node (URI or bnode) with such a 
graph.  This approach is sometimes called "contexts", though that term may 
sometimes be over-constraining.

(3) The Harmony project introduced an idea called ABC [3], a variant of 
which has been picked up by some Dublin Core work [4], which allows one to 
create a refined version of a property by creating a composed relationship 
that indirects via an intervening node.  This doesn't directly address the 
question you ask, but your mention of "there are/could be more refinements 
of the property dc:relation" hints that this might be what you are really 
trying to do.


[1] http://www.w3.org/TR/rdf-mt/


(on named graphs, provenance and trust (Jeremy Carroll, Chris Bizer, Pat 
Hayes and Patrick Stickler).

[3] http://www.ilrt.bris.ac.uk/discovery/harmony/docs/abc/abc_draft.html
(cf. section 3.4)
     http://metadata.net/harmony/JODI_Final.pdf  (I think this is relevant)

[4] Expressing Qualified Dublin Core in RDF / XML:

At 10:27 20/05/04 +0100, Richard Lennox wrote:
>I was wondering if this is at all possible.
>I have a triple:
>Class1 dc:relation  Class2
>Such that there is a relation between Class1 and Class2 - and while there 
>are/could be more refinements
>of the property dc:relation, say for the sake of arguments they are not 
>correct in this particular usage.  Is it then possible to add a property 
>to the property, a particular comment.
>Class1 dc:relation Class2
>dc:relation comment blahblah
>Perhaps those triples are wrong, I just don't know.
>And how would I reprresent this in various manifestations of RDF (N3, 
>(R,P,V) and RDF/XML)?
>Many thnaks
>Richard Lennox

Graham Klyne
For email:
Received on Thursday, 20 May 2004 06:23:24 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:07:51 UTC