Re: About rdfs:member and rdf:_nnn

>Comments on rdfs:member and rdf:_nnn.
>
>1. About rdfs:member
>
>Why not constraint the rdfs:domain of rdfs:member to be rdfs:Container?
>    (3.3 RDFS interpretations of the RDF Semantics spec)

We could, indeed. This would constrain any other uses of rdfs:member 
on user-defined container types, however, and it does not seem that 
it would provide very much useful functionality. I will however raise 
this matter with the WG.

>     Consider adding an axiomatic triple as follows:
>     rdfs:member rdfs:domain rdfs:Container
>
>Why not specify rdfs:member to be an instance of 
>rdfs:ContainerMembershipProperty?
>    (3.3 RDFS interpretations of the RDF Semantics spec)
>
>     Consider adding an axiomatic triple as follows:
>     rdfs:member rdf:type rdfs:ContainerMembershipProperty
>
>By so doing, we have:
>* rdfs:member is an instance of rdfs:ContainerMembershipProperty 
>(not just rdf:Property) that is a super-property of all the 
>container membership properties.

Right, that is an option. However there is also the intuition that 
the ContainerMembershipProperties are totally ordered, and 
rdfs:member is nowhere in that ordering. Again I think this is best 
discussed by the WG.

Brian, an issue number for this??


>* Container membership properties must be applied to containers.
>
>These reflects most of the intended meanings or use cases. Of 
>course, the current treatment is not wrong. (See section 5.1.5, 
>section 5.1.6  and Appendix A of the RDF Schema Spec, section 3.3 of 
>the RDF Semantics spec)
>
>2. About rdf:_nnn  (section 3.2.2 of the RDF Semantics spec)
>
>Please consider adding one more example to clarify the meaning of 
>rdf:_nnn, such as the following:
>
>_:xxx rdf:type rdf:Seq.
>_:xxx rdf:_2 <ex:a> .
>_:xxx rdf:_2 <ex:c> .
>
><ex:a> and <ex:c> should denote the same thing? In other words, the 
>property rdf:_nnn should be a functional property? (I haven't found 
>any explanation about this issue in the spec)

Since RDFS has no notion of equality, there is no way to express the 
idea of a functional property in RDFS. Whether or not a property is 
functional can make no difference to any RDFS entailment. In semantic 
extensions like OWL which can express the idea of a functional 
property, it would certainly be natural to impose this requirement; 
but then that must be part of the spec of the extended language.

I have an editorial problem with this section, to be honest. On the 
one hand, it might be helpful to list all the non-entailments that 
anyone might ever think of. On the other hand, these sections of the 
semantics document are already considered to be too long by many 
readers and some have urged that they be deleted. I will try to find 
a way of making the point clear to the reader without unduly 
lengthening the section.

>Considering there are two RDF graphs, the first one contains:
>
><foo:aContainer> rdf:type rdfs:Container.
><foo:aContainer> rdf:_1 <ex:a> .
><foo:aContainer> rdf:_2 <ex:b> .
>
>The second one contains:
><foo:aContainer> rdf:_1 <ex:b> .
><foo:aContainer> rdf:_2 <ex:a> .
>
>How about merging the two RDF graphs?

Merging these graphs will result in a graph which makes some rather 
odd assertions about the container. This is not strictly speaking 
inconsistent in RDFS, but it does violate what might be called the 
intended mode of use, particularly if the container is a sequence for 
example. In an appropriate semantic extension, one would conclude 
from the merge that <ex:b> must be identical to <ex:a>, for example 
(I believe that would be a valid conclusion in OWL, although I have 
not checked the details recently.)

Pat Hayes

>
>Thanks for your concern!
>
>
>Yuzhong Qu
>--------------------------------------------------
>Dept. of Computer Science and Engineering
>Southeast University, Nanjing 210096, P. R. China
>Home Page: http://cse.seu.edu.cn/People/yzqu/en
>Research Group: http://xobjects.seu.edu.cn
>--------------------------------------------------


-- 
---------------------------------------------------------------------
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@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
s.pam@ai.uwf.edu   for spam

Received on Tuesday, 18 February 2003 12:25:08 UTC