Some comments prompted by the semantics document

I've been thinking about implementing inferences based on the axioms and 
closures in the semantics document [1], and various perceptions have surfaced:

1.  Why rdfs:Container?   It seems a new vocabulary term has been 
introduced, but it doesn't seem to serve any real purpose.  In particular 
it doesn't figure in the domain of rdf:_n or rdfs:member.

2.  Was there a reason for not specifying rdfs:member to be an instance of 
rdfs:ContainerMembershipProperty?  I vaguely recall some discussion of 
this, but can't remember the substance.

If there are particular reasons for (1) and (2) above, I think it would be 
good to add a note to the semantics document, as it's very easy to think 
these might be accidental omissions.


3. Appendix B, "entailment rules".

(I just noticed, is it really correct to call these "entailment rules", 
since "entailment" is a semantic condition and these are more like proof 
rules (i.e. syntactic in character).

My main suggestion was that it would be good to add cross-referencesfrom 
here to section 3, where the axiomatic triples are listed.  (Preferably one 
for each of the RDF- and RDFS- closure rules tables.)

4.  I found myself wondering about the purpose of the RDF axiomatic triples 
in section 3, all but one of which simply assert that the rdf property 
terms are of type rdf:Property.  But as soon as one uses them in this way, 
this fact follows from the closure rule rdf1.  I recognize these are not 
logically congruous statement, but I can't see any useful result from 
knowing a-priori that, say, rdf:subject has type rdf:Property.

#g
--

[1] RDF semantics, 18 Jun 2003 editor's revision
(No internet connection:  can't get online to retrieve URI.)


-------------------
Graham Klyne
<GK@NineByNine.org>
PGP: 0FAA 69FF C083 000B A2E9  A131 01B9 1C7A DBCA CB5E

Received on Wednesday, 25 June 2003 19:28:20 UTC