Re: Case for Reinstatement of Qualified Cardinality Restrictions

Peter,

Guus's comment got me thinking. My original message was based on what
was true for DAML+OIL, and I just assumed it wasn't changed in OWL
because I didn't remember it coming up as an issue. However, I started
poking around in the AS&S section 4, and don't see a transformation from
the abstract syntax that would produce a piece of RDF syntax in which an
owl:Restriction had a combination of owl:cardinality, owl:allValuesFrom,
owl:someValuesFrom, etc. The only way this could happen is if different
restrictions could have the same main node, but I'm assuming that the 
"_:x" notation means a blank node, which you say must be local to each
transformation. If the RDF syntax of OWL is only those RDF graphs that
can be produced from abtract OWL ontologies by applying the
transformation rules, then it seems that Guus is right (if I'm wrong
please point out how the transformation can be performed or where my
assumptions went astray).

In any case, this exercise has led me to believe we need to do two
things with the OWL specs. First, I think the reference needs to discuss
what is the preferred way for writing multiple restrictions on a single
property (I assume this is as an intersectionOf Restrictions, although
this is very ugly, and I much prefer the syntax used in my previous
message, if it's legal and correct OWL).

Second, although the AS&S makes it easy to transform from the abstract
syntax to RDF, I think it is much harder to do the reverse
transformation (which actually seems like a more common one). I think it
would be nice to have a table that presents this transformation. This
may also be useful in placating people who ask "where's the official
syntax?"

Jeff

"Peter F. Patel-Schneider" wrote:
> 
> From: Guus Schreiber <schreiber@cs.vu.nl>
> Subject: Re: Case for Reinstatement of Qualified Cardinality Restrictions
> Date: Wed, 23 Apr 2003 11:08:22 +0100
> 
> >
> > Jeff Heflin wrote:
> >
> > [..]
> >
> > > 3) I think the factor that makes QCRs most confusing in OWL is the
> > > difficulty in expressing them cleanly in triples. Currently, the
> > > Restriction class is a place to hang each restriction that is applicable
> > > to a property. Currently, each of these is a binary predicate so the
> > > following is perfectly fine.
> > >
> > > <owl:Restriction>
> > >   <owl:onProperty rdf:resource="&eg;hasDigit"/>
> > >   <owl:cardinality rdf:datatype="&xsd;int">5</owl:cardinality>
> > >   <owl:allValuesFrom rdf:resource="&eg;Digit" />
> > >   <owl:someValuesFrom rdf:resource="&eg;Finger" />
> > > </owl:Restriction>
> > >
> >
> > This is not "perfectly fine". OWL currently allows only one single value
> > or cardinality constraint witrhin a restriction class (se the relevant
> > sections in S&AS and Ref).
> 
> Huh?  OWL allows an arbitrary number of restriction triples (onProperty,
> cardinality, etc.) to have the same subject.  Sometimes OWL gives no extra
> meaning to these restrictions; sometimes OWL gives the extra meaning that
> one would expect; sometimes OWL gives unusal extra meaning to these
> restrictions.  In particular, Jeff's example is one case where the meaning
> given to a restriction by OWL is decidedly unusal.
> 
> It would have been much nicer if this was not the case, but our hands were
> tied.
> 
> [...]
> 
> > Guus
> >
> > [..]
> 
> peter

Received on Wednesday, 23 April 2003 14:29:22 UTC