- From: Brian McBride <bwm@hplb.hpl.hp.com>
- Date: Wed, 28 Nov 2001 17:50:45 +0000
- To: www-rdf-interest@w3.org
Hi Marc, You prompted elsewhere to reply to this. I'm now on the train back. At 09:19 21/11/2001 +0000, you wrote: [...] So, I say that a person has a car or a motorcycle, or both or them, ok? Ok'ish. You have made a closed world assumption, i.e. that nothing else can be in the domain of the property, which is fine for the purposes of validation. [...] With the new model I should 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#Vehicle"/> </rdf:Property> Later, in another example your propose to use CarsAndMotos. Why here have you not proposed to use CarsOrMotos? Now if you did, we would have Cars are a subclass of CarsOrMotos, Motos are a subclass of CarsOrMotos - ah - I see what you mean, even with the CWA, we can't be sure that there are no other members of CarsAndMotos. But so what? You are doing this for validation, right? So if you have foo, a subject for this property you want to check that you know its type satisfies the domain contraint. If it has type Car, all is well; if it has type Moto all is well. If it has type anything else, then unless that type is defined to be a subclass of CarsAndMotos, or is CarsAndMotos you're validator is going to baulk - right. And that is never going to happen. So your validator will fail anything that is the subject of the property which is not known to be a car or a moto, and that is what you wanted. Brian
Received on Thursday, 29 November 2001 05:39:09 UTC