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

Re: Placing a comment on a relationship?!

From: Thomas B. Passin <tpassin@comcast.net>
Date: Thu, 20 May 2004 15:05:33 -0400
Message-ID: <40AD017D.4000006@comcast.net>
To: www-rdf-interest@w3.org

Richard Lennox wrote:

>>Either way, you would only use a given one of these predicates once in a
>>given graph.
> Not necessarily a single thing can have relations to many things and using
> many different types of relation hence a graph could have
> C1 dc:relation_1 C2
> C1 dc:relation_2 C3
> C1 dc:relation_3 C4
> Defining all subProperties of all possible relations would not be possible
> and expecting non-tehcnical users to create their own "subPropertiesOf"
> would probably be unfeasible.

That's not the point here, IMHO.  If you want to make a comment about 
one particular statement, you have to deal with a single instance of 
_something_.  Making that something be a single instance of a predicate 
is one possible way.  Without a native ability to refer to a specific 
triple (or collection of triples) as a resource in itself, any 
workaround is bound to be clumsy or unpleasant.

If you look at large-scale, working systems organized using binary 
relations, like Metis, you often find that the relations may contain a 
great deal of meta data.  That's generally needed for practical reasons, 
like having an audit train or specifying the line density and style. 
But we can't attach meta data to a specific relationship in RDF unless 
we treat it as an instance.  Since all predicates are also resources, 
this is possible in principle.

Now how can we get a specific predicate to be an instance of a property? 
  One way (which just simulates the desired construction) is to declare 
it to be a subProperty of its parent class, and then use it for just one 
triple.  No other constaints need be applied, except that we had better 
not reuse it for more than one triple.

Another way would be to make it an instance by declaring that it is a 
_type_ of its parent property.

> Perhaps the intended triple is:
> (c1 dc:relation C2) x:comment "blah blah"  .
>>From what I understand reification is a way to do it but it is very verbose
> (especially in RDF/XML)

That is sort of the intent, but a reified statement does not actually 
refer to any particular triple, so making a comment about the reified 
statement isn't quite the same thing as making a comment about a 
specific triple.

Besides, using reification is more verbose than using a property 
instance or a unique subproperty, I think.


Tom P
Received on Thursday, 20 May 2004 15:08:10 UTC

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