Re: Domain/Range: conjuntion or disjuntion??

From: tarod@softhome.net
Subject: Domain/Range: conjuntion or disjuntion??
Date: Wed, 21 Nov 2001 09:19:41 GMT

> 
>   Hello Brian,
> 
>   With the old model, If I wanted to say that a property has as range a
> class that must be a Car and a MotorCycle I could add to the schema.
> 	<rdfs:Class rdf:about="http://some.com#CarsAndMotos">
> 		<rdfs:subClassOf rdf:resource="http://some.com#Car"/>
> 		<rdfs:subClassOf rdf:resource="http://some.com#MotorCycle"/>
> 	</rdfs:Class>
>   and change 1 for	
> 	<rdf:Property rdf:about="http://some.com#owns">
> 		<rdfs:domain rdf:resource="http://some.com#Person"/>
> 		<rdfs:range rdf:resource="http://some.com#CarsAndMotos"/>
> 	</rdf:Property>
> 	
>   now I can only have CarsAndMotos as range of the property own and because
> 
>   SubClass(CarsAndMotos, Car) and SubClass(CarsAndMotos, MotorCycle);
>   Range(own, CarsAndMotos);
> 
>   Range(x, y) and SubClass(z, y) -> Range(x, z);
>   
>   Range(own, Car) and Range(own, MotorCycle) !!!!!! THAT's what you wanted,
> isn't it?

No.  CarsAndMotos is not the same as the intersection of Car and
Motorcycles.  It is possible to have an object that is both a Car and a
Motorcycle, but not a CarsAndMotos.

So, in some sense, and remembering the expressive impoverishment of RDFS,
the expressive power of the disjunctive view and the expressive power of
the conjunctive view are incomparable.  There are tricks that can be played
in RDFS to get part of the difference back, but the difference is only
eliminated if you have true disjunction and conjunction as concept-forming
operators.

Peter F. Patel-Schneider
Bell Labs Research

Received on Wednesday, 21 November 2001 09:03:17 UTC