RE: Restriction

On August 7, Gary Ng writes:
> 
> Thanks Peter, 
> 
> It is probably a bad example, but I was simply trying to state a general
> axiom:
> 
> 	exists callThemselves "Human" implies HomoSapien.
> 
> Which is exactly what you said in your PS
> 
> > > It does not violate the definition of owl restriction, as it has
> exactly
> > > one onProperty, and exactly one value constraint.
> > ...
> > means.  However, it is not legal OWL DL, because there is no way to
> have
> > extras properties on restrictions in OWL DL.
> >
> Hmm... ok, I can accept that. Just so that I could reference it in my
> code, I would be grateful if you could kindly point me to the location
> where this is explicitly mentioned. I have been trying to locate this in
> the specification docs (last call), I am afraid I cannot find any
> supporting evidence in the doc that indicates it is invalid OWL-DL.
> 
> According to the reference doc, 
> 
> 	- a restriction is a special kind of class description, and
> 	- AXIOM SCHEMA: class description rfs:subClassOf class
> description
> 
> And the AS&S seems to support these two statements too.

The abstract syntax does not allow for more than one property in a
restriction. The definition of valid RDF syntax is that it must be
possible to generate it from abstract syntax using the transformation
rules. It would not be possible to generate your example.


> 
> Given that I cannot use this RDF form to say it in OWL-DL, then I'd have
> to use:
> 
> <owl:Class>
>   <owl:equivalentClass>
>     <owl:Restriction>
>       <owl:onProperty rdf:resource="#callThemselves" />
>       <owl:hasValue   rdf:datatype="&xsd;string">Human</owl:hasValue>
>     </owl:Restriction>
>   </owl:equivalentClass>
>   <rdfs:subClassOf rdf:resource="#HomoSapien"/>
> </owl:Class>
> 
> or 
> 
> <owl:Class>
>   <owl:intersectionOf> <!-- or unionOf... shouldn't make any difference
> -->
>     <owl:Restriction>
>       <owl:onProperty rdf:resource="#callThemselves" />
>       <owl:hasValue   rdf:datatype="&xsd;string">Human</owl:hasValue>
>     </owl:Restriction>
>   </owl:intersectionOf>
>   <rdfs:subClassOf rdf:resource="#HomoSapien"/>
> </owl:Class>
> 
> All three forms should result in the same semantics right? Although the
> exact number of axioms and their structure may be slightly different.

As an owl restriction is a class, you can simply state the subClass
axiom/property directly w.r.t. the restriction:

    <owl:Restriction>
        <owl:onProperty rdf:resource="#callThemselves"/>
        <owl:hasValue rdf:datatype="&xsd;string">Human</owl:hasValue>
        <rdfs:subClassOf>
            <owl:Class rdf:about="#HomoSapiens"/>
        </rdfs:subClassOf>
    </owl:Restriction>

Regards, Ian


> 
> Many thanks, 
> 
> Gary
> 
> > -----Original Message-----
> > From: Peter F. Patel-Schneider [mailto:pfps@research.bell-labs.com]
> > Sent: 06 August 2003 18:02
> > To: Gary Ng
> > Cc: www-rdf-logic@w3.org
> > Subject: Re: Restriction
> > 
> > From: "Gary Ng" <Gary.Ng@networkinference.com>
> > Subject: owl:Restriction
> > Date: Tue, 5 Aug 2003 12:04:48 +0100
> > 
> > >
> > >
> > > Hi all,
> > >
> > > I am confused as to whether the following piece of OWL is valid:
> > >
> > > <owl:Restriction>
> > >   <owl:onProperty rdf:resource="#callThemselves" />
> > >   <owl:hasValue   rdf:datatype="&xsd;string">Human</owl:hasValue>
> > >   <rdfs:subClassOf rdf:resource="#HomoSapien"/>
> > > </owl:Restriction>
> > >
> > > It does not violate the definition of owl restriction, as it has
> exactly
> > > one onProperty, and exactly one value constraint.
> > >
> > > I believe this is valid (albeit unconventional) as owl:Restriction
> is a
> > > "subclass" of owl:Class. The normative RDF Schema for owl seems to
> > > suggest that anything that can appear within owl:Class can certainly
> > > appear in owl:Restriction in the RDF XML format. (e.g. the
> rdfs:domain
> > > for intersectionOf is a Class).
> > >
> > > Please yell if I am totally wrong. If it is invalid, we need to make
> > > sure our tools do not export this form.
> > >
> > > Cheers,
> > >
> > > Gary
> > >
> > >
> > > Gary Ng, PhD 	     gary.ng@networkinference.com
> > > Senior Software Engineer
> > > Network Inference (Holdings) Ltd
> > > Tel: +44 (0) 20 7616 0717 Fax: +44 (0) 20 7616 0701
> > 
> > 
> > This is legal OWL Full, and might even mean what you think that it
> > means.  However, it is not legal OWL DL, because there is no way to
> have
> > extras properties on restrictions in OWL DL.
> > 
> > peter
> > 
> > PS:  The meaning of this is that if something has a callThemselves
> > property value of "Human" then it is a HomoSapien.

Received on Saturday, 9 August 2003 20:56:07 UTC