Re: Domain of rdfs:member

>According to the latest RDF schema document, the domain of the 
>rdfs:member property
>(which is inherited to all rdf:_nnn properties) is class rdfs:Container.
>However, it has been mentioned many times that rdf:_nnn properties 
>can be "attached"
>to any resource, not only members of the rdfs:Container class.

Well, the semantic rule says that if you attach an rdf:_nnn property 
to something, then that thing *is* a container. That is, the 
definition of a container is that it is something that these 
properties are applied to.

>Furthermore, many test-cases concerning the 
>rdf-containers-syntax-vs-schema issue
>demonstrate the above fact, that rdf:_nnn properties are attached to 
>any resource.
>I believe that this is contradictory.

Formally it is not a contradiction. See below.

>Should a resource with an rdf:_nnn property be considered a member 
>of rdfs:Container class,
>even if it has been described as a member of another class?

Yes. In general, entities can  be members of more than one class. So 
for example something could be in the class of People and also the 
class of Containers. While this might seem odd, there is no way in 
RDFS of asserting that two classes are disjoint (in general, RDFS 
provides no way to express 'negative' information of this kind), so 
such an assertion cannot produce a contradiction.  More expressive 
languages which can express disjointness could of course forbid such 
combinations, but to find those meanings one would need to consult 
the semantic rules of those extended languages.

Pat Hayes
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax

Received on Monday, 16 September 2002 16:41:34 UTC