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

Re: Challenge for RDF Gurus :)

From: <tarod@softhome.net>
Date: Thu, 14 Feb 2002 20:15:48 GMT
Message-ID: <20020214201548.25199.qmail@softhome.net>
To: "Seth Russell" <seth@robustai.net>
Cc: www-rdf-logic@w3.org, www-rdf-comments@w3.org, www-rdf-interest@w3.org

Seth Russell writes:

> From: <tarod@softhome.net>
> 
> > Seth Russell writes:
> >
> > > From: <tarod@softhome.net>
> > >
> > > > > re: http://robustai.net/mentography/rdfs_domain_range2.gif
> > > > >
> > > > > >   Good try but I must say that it's not 100% what I asked for
> because
> > > for
> > > > > > the range issue you use
> > > > > >   Class C
> > > > > >   A is subClassOf C
> > > > > >   B is subClassOf C
> > > > > >   And then c range is C. It's a good aproach but it's not
> logically
> > > > > > correct, you are saying that range of c is (C or A or B) and I
> asked
> > > for
> > > > > > range of c should be (A or B)
> > > > >
> > > > > Ok, I saw this problem after I published the graph.  I would need a
> way
> > > to
> > > > > say that there is no instances of C which is not and instance of A
> or B.
> > > > > I'm beginning to agree with Sean, there is no way to say this with
> the
> > > > > primitives of rdfs only.
> > > >   It was posible before some RDFCore changes :)
> > > >
> > > > > What is your objection to using the daml schema?
> > > >
> > > >   I have no objection, this is just a challenge.
> > > >
> > > > > >   Now try it with the old aproach it's easier.
> > > > >
> > > > > What approach are you talking about here?
> > > >
> > > >   Before some changes in the schema, that a property had two domains
> (at
> > > > the begining a property must only have one range, now it can have more
> > > than
> > > > one) means that the subject of the property must be in one of those
> > > > domains, it was a disjuntion of restrictions. When they added more
> than
> > > one
> > > > range if they had used this vision, the value of a property must be a
> > > > member of one of the domains, if that make sense to you, try it now.
> It's
> > > > very easy having this in mind.
> > >
> > > Well if the domain restraint is jisunctive and the range restraint is
> > > conjunctive, then I suppose your example would eaisly work that old way.
> > > But if one wanted the opposite case ( range (A and B).  domain ( A or
> B)),
> > > then we still couldn't do it.   Intiitively don't we want domain and
> range
> > > to be symmetric here?
> > >
> > > Mentograph available upon request.
> > > Seth Russell
> >
> >   Sorry, I meant both of them should be disjuntive.
> 
> Ok, I think I see what you meant.  Is this it?
> 
> http://robustai.net/mentography/rdfs_domain_range4.gif
> 
> But then wouldn't that would break one of the basic concepts of RDF grpahs
> ... all triples are anded together ?
> 
> Seth Russell
> 
> 

  I donīt understand what do you mean with "all the triples are anded
togheter"

  But you model is so close what I wanted :)
  c only has one domain (C) (and all instances of C are instances of A and
B), you added another domain and I don't know why. And two ranges, Class A
and Class B, so the range is A or B. So easy!!!!! And you don't need to add
the DisjuntionProperty, it's not necesary.

  Thanks,
         Marc
Received on Thursday, 14 February 2002 15:12:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:42 GMT