W3C home > Mailing lists > Public > public-vocabs@w3.org > March 2014

Re: How to avoid that collections "break" relationships

From: Peter F. Patel-Schneider <pfpschneider@gmail.com>
Date: Mon, 31 Mar 2014 13:09:55 -0700
Message-ID: <5339CB93.5030805@gmail.com>
To: Markus Lanthaler <markus.lanthaler@gmx.net>, public-hydra@w3.org, 'Linked Data community' <public-lod@w3.org>, 'W3C Web Schemas Task Force' <public-vocabs@w3.org>

On 03/31/2014 11:35 AM, Markus Lanthaler wrote:
> On Monday, March 31, 2014 7:09 PM, Peter F. Patel-Schneider wrote:
>> In actuality, defining things like owl:sameAs is indeed extending RDF.
>> Defining things in terms of OWL connectives also goes beyond RDF. This
>> is different from introducing domain predicates like foaf:friends.
>> (Yes, it is sometimes a bit hard to figure out which side of the line
>> one is on.)
> Peter, could you please describe where you draw the line? I generally look at RDF as a data model and in those terms nothing that has been suggested really extends RDF. You probably have something related to semantics in mind, please note however that Ruben (please correct me if I'm wrong) didn't suggest to change RDF's semantics, i.e., the triples
>
>     </people/markus> foaf:knows [
>         hydra:memberOf </people/markus/friends>
>     ] .
>
> enable no inferences to be made on /people/markus/friends. More specifically, they do not say that the members of /people/markus/friends are related to /people/markus via foaf:knows.
>
> But perhaps you had something completely different in mind!?
>
>
> --
> Markus Lanthaler
> @markuslanthaler
>
>

Well, if you view RDF as simply a data model there is no line to be drawn.  
However, then you are stuck with RDF as a data model - there are no inferences 
to be made and so nothing to break.

However, if you view RDF as representing information according to the 
definition of RDF, and particularly if you are actually using RDFS, then there 
are inferences to be drawn when using particular constructs.  So, for example, 
if you say
foo:bar rdfs:subPropertyOf rdfs:subClassOf .
foo:student foo:bar foo:person .
foo:john rdf:type foo:student.
then you have implicitly also said
foo:john rdf:type foo:person.

RDFS licenses a certain set of these inferences, and no more.

If you want to add to these inferences, then you are crossing the line.  Using 
foaf:friend doesn't add inferences, and so doesn't cross the line.  On the 
other hand, the use of owl:sameAs generally comes with added inferences, and 
so does cross the line.

Similarly, the use of hydra:memberOf in the proposed manner is precisely to 
add in new inferences.  The point is that RDFS processors will not make these 
inferences and so the proposal is indeed extending RDFS.

peter
Received on Monday, 31 March 2014 20:10:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:38 UTC