W3C home > Mailing lists > Public > www-rdf-comments@w3.org > October to December 2003

Re: Some comments about the relationships between rdfs:subPropertyOf and rdfs:domain(rdfs:range)

From: pat hayes <phayes@ihmc.us>
Date: Wed, 3 Dec 2003 10:54:28 -0600
Message-Id: <p06001f1bbbf3c06e9d2e@[10.1.31.1]>
To: "lsp" <lsp@is.pku.edu.cn>
Cc: www-rdf-comments@w3.org
>Hi,
>

Hi, sorry I missed this message earlier. You may have already had 
replies, but in case not:

>I have some comments about rdfs:subPropertyOf:
>1) if  (R rdfs:subPropertyOf S)
>      (R rdfs:domain C1)  (R rdf:range D1)
>      (S rdfs:domain C2)  (S rdf:range D2)
>   can we infer that  (C1 rdfs:subClassOf C2) && (D1 rdfs:subClassOf D2)?
>

No. What you can infer is that if x R y then x rdf:type C2 and y 
rdf:type D2, for any x and y; but being a subSET does not 
automatically make it a subCLASS.

>I believe the result can be easily inferred by the RDF Model Theory,

No, it does not. The semantic conditions on subClassOf are phrased in 
if-then terms, not iff. See
http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-mt-20030117/#rdfsinterpdef 
, 4th and 5th lines of the table.

Are you looking at a recent version of the MT draft? This was changed 
about 6 months ago. The LC version has a discussion of this issue, 
under the heading 'extensional semantic conditions', see

http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-mt-20030117/#ExtensionalDomRang

In this respect RDFS is weaker than OWL, for example, which does 
impose extensional conditions and in which the inference you describe 
would be valid. We took the decision to keep RDFS as simple as 
possible while still supporting most 'ordinary' uses of subClass and 
subProperty: it is the family car, not the high-performance vehicle.

>but there is no rule in the RDFS closure rules
>to show this inference.
>
>2) rdf schema allows one rdf:Property has multiple rdfs:domain and 
>rdfs:range,  and  the resulted domain and range is
>considered as intersection.

Right. Or at least, a subclass of the intersection, to be exact.

>But RDF schema specification didn't say much what about have zero 
>rdfs:domain or rdfs:range,
>this arise some confusion.

? Do you mean, if no domain or range is specified? In that case you 
simply have no knowledge of the true domain and range.

>a)       is zero rdfs:domain or rdfs:range declaration is legit in rdf schema?

It is legit to omit anything, including domain and range information.

>b)       If zero rdfs:domain declaration means that the property's 
>domain is rdfs:Resource,

It does not mean that. True, the domain is a subclass of 
rdfs:Resource, but then all classes are.

>there is a problem: if the property's superproperty
>'s domain is declared as Class C1, then we can conclude the 
>propery's domain is a subclass of C1, but not rdfs:Resource.

I do not see what the problem would be here, but never mind.

>c)       What is the meaning of zero rdfs:range declaration?

Im not sure what you mean by 'zero'. If you mean, what is the meaning 
of there being no domain specified, then the answer is, the meaning 
is that the domain is not specified. That should not be understood as 
saying that there is no domain, only that your knowledge of it is 
incomplete. In RDF/S, one assumes that knowledge may always be 
incomplete.

Pat Hayes

>
>Thanks!
>
>Liu shengping
>


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes
Received on Wednesday, 3 December 2003 11:57:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 21 September 2012 14:16:33 GMT