W3C home > Mailing lists > Public > public-rdf@w3.org > March 2012

Re: Can't RDF describe collection resources?

From: Ruben Verborgh <ruben.verborgh@ugent.be>
Date: Thu, 1 Mar 2012 10:48:05 +0100
Cc: public-rdf@w3.org, public-rdf-wg@w3.org, W3C SWIG Mailing-List <semantic-web@w3.org>
Message-Id: <BF73B23D-2666-407A-AA3F-2AE03C3CEFDD@ugent.be>
To: Enrico Franconi <franconi@inf.unibz.it>
Hi Enrico,

>> Plus, I would indeed need to make the relation between plural and singular explicit for each property with OWL property chains. (But at least, itís possible.)
> I don't get this: you just need one property chain axiom, saying that :hasComment is the composition of :hasComments (as a functional property) with :hasMember.

The thing is that Iím trying to solve the general problem here.
For the particular instance of :hasComment <=> :hasComments, you indeed need one axiom and thatís straightforward.
However, to make this work for all properties (foaf:knows, rdf:labelÖ) Iíd have to make an extra property and an extra axiom for all of them.

>> Oh, of course, I see it now.
>> The issue is that RDF in practice always works the other way round: it tends to describe individuals instead of collections.
> I don't get this. Remember that with this modelling you would have at your disposal both :hasComments and :hasComment, being them redundantly constrained by the property chain axiom. You can use any of the two interchangeably. In aprticular, you can use just :hasComment and ignore systematically :hasComments in your application (but when you need to create the comments resource once forever for a newly created blog) and everything would be fine. To be more precise, you need also to state that the :hasComments property is functional, so that you could add (or refer to) a comment as a :hasComment without bothering mentioning the comments resource.

If I understand correctly, you argue that the :hasComment property is redundant, because it is equivalent to the composition of :hasComments and :memberOf.
(The :hasComments is then not redundant, for cases where the comment set is empty.)
So what I meant, is that the majority of existing data (and properties) have been described with the singular properties, because thatís just easier in RDF.


Received on Thursday, 1 March 2012 09:48:46 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:37:11 UTC