Re: subclasses (RDF vocabulary definitions)

Richard H. McCullough wrote:

> In enumerating the alternatives, I am NOT talking about
> 
>     aa rdf:type Animal  entails  aa rdf:type Man
>  
> I am talking about
> 
>     Man  rdfs:sameAs  Animal


I'll try again:


I know that's what you're explicitly talking about;  but "sameAs" is 
just a piece of syntax unless we can also talk about what "sameAs" 
*means*.  And it seems to me we have to talk what "sameAs" means by 
looking at the logical consequences of stating it.  I'm saying that if 
you say

Man "sameAs" Animal, you are saying

if aa is a Man, then aa is an Animal

  

AND

if aa is an Animal, then aa is a Man.

Putting that in RDF entailment rule form (which is what I did earlier in 
this thread for the entailments allowed by rdfs:subClassOf), Man 
"sameAs" Animal would mean

Man sameAs Animal
aaa rdf:type Man

entails

aaa rdf:type Animal

and

Man sameAs Animal
aaa rdf:type Animal

entails

aaa rdf:type Man

On the one hand, I take it you will agree that, at any point in time, it 
might very well be the case that the set of resources (the set of things 
with URIs) that are also described as being of class Animal (having 
rdf:type properties with class Animal as the value) might very well be 
the same as the set of resources described as being of class Man (having 
rdf:type properties with class Man as the value), even if you've stated 
that Man is a proper subset of Animal?  The two sets being the same is 
an "accident" of sorts:  I've just not described any wolves or tigers 
yet.  But this situation surely must be allowed to exist.  On the other 
hand, if I have the "sameAs" *definition*, this is the statement of a 
*rule* I should be able to make inferences from:  I should be able to 
infer from the fact that something is an Animal that it necessarily must 
be a Man, and vice-versa.

Now, Man rdfs:subClassOf Animal certainly allows for the situation above 
to exist where the sets of resources you've so-far identified as Man and 
Animal happen to be the same.  However, my point is that, if 
rdfs:subClassOf, *as a definition*, did not rule out the possibility 
that the two class had to be the same, it would allow both the "sameAs" 
entailments:

Man rdfs:subClassOf Animal
aaa rdf:type Man

entails

aaa rdf:type Animal

and

Man rdfs:subClassOf Animal
aaa rdf:type Animal

entails

aaa rdf:type Man

And my point was that RDF does not license this second entailment.


--Frank



-- 
Frank Manola                   The MITRE Corporation
202 Burlington Road, MS A345   Bedford, MA 01730-1420
mailto:fmanola@mitre.org       voice: 781-271-8147   FAX: 781-271-875

Received on Friday, 22 November 2002 14:39:50 UTC