W3C home > Mailing lists > Public > www-rdf-logic@w3.org > February 2002

Re: reification test case

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Thu, 07 Feb 2002 11:07:56 -0500
To: seth@robustai.net
Cc: www-rdf-logic@w3.org
Message-Id: <20020207110756F.pfps@research.bell-labs.com>
From: "Seth Russell" <seth@robustai.net>
Subject: Re: reification test case
Date: Thu, 7 Feb 2002 06:31:13 -0800

> From: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
> 
> > > RDF is used to describe resources, right?   Bnodes describe resources.
> The
> > > use of the word 'single' is what we are arguing about.  Certainly Bnodes
> do
> > > not necessarily describe only a *single* object in the domain of our
> > > discourse.  One Bnode can describe many objects at once. For example:
> the
> > > following RDF Bnode description, written in N3 with it's corresponding
> > > schema (not included), certainly does not describe only one *single*
> > > automobile.
> > >
> > >    [rdf:type :Automobile;
> > >     :has :wheels, :motor; :
> > >     :manufacturedBy :GM;
> > >     :modelName "Oldsmobile"]
> >
> > No, no, no, a thousand times no!!!!
> 
> Ok, ok, ok .. I was wrong .. mea copa.   So how do we represent the class of
> oldsmobiles?
> 
>     [rdf:type rdfs:Class;
>      :has :wheels, :motor; :
>      :manufacturedBy :GM;
>      :modelName "Oldsmobile"]

Well, not really.  The class of Oldsmobiles is not manufactured by GM, for
example.  To represent the class of Oldsmobiles you have to go beyond
RDF(S) to something like OIL.  You seem to be wanting to say something
like:

	{ x : manufacturedBy(x,GM) ^ modelName(x,"Oldsmobile") }

> But then can we say the following?
> 
>   [rdfs:subclass  rdf:Statement
>    rdf:subject  :S1
>    rdf:predicate :P1
>    rdf:object :O1
>    dc:author :Seth]
>
> If not, why not?

Sure you can say this, but this is not the class of statements that have
subject :S1.  Instead this is a class of reified statements and, as well,
an object that is the subj of a statement with pred rdf:subject and obj
:S1, ....

Here you are exhibiting the canonical type/token confusion.  Properties of
a class are not properties of objects that belong to the class, and vice
versa.  To state these things about the objects that belong to a class you
need facilities like those provided by a description logic.

> Seth Russell

Peter F. Patel-Schneider
Received on Thursday, 7 February 2002 11:09:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:41 GMT