Re: Properties no longer required to be resources?

>I just noticed an editorial (?) tweak (pointed out by pfps) to the
>semantics document that the semantic constraint that properties must be
>a subset of resources has been removed from the current editors draft.
>
>http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-mt-20030117/#interp
>
>Pat, this is rather more than just an editorial tweak.

This is only for simple interpretations. It does not make any 
difference to RDF or RDFS interpretations, since those 
interpretations are still required to conform to the subset condition 
by virtue of their semantic conditions; cf. section 3, 5th para 
(after the table of RDF semantic conditions):

"The first condition could be regarded as defining IP to be the set 
of resources in the universe of the interpretation which have the 
value I(rdf:Property) of the property I(rdf:type). Such subsets of 
the universe will be central in interpretations of RDFS. Note that 
this condition requires IP to be a subset of IR."

So all of the following is still true in RDF, and in all RDF semantic 
extensions.

>1. M&S states that properties are a subset of resources
>2. Object oriented implementations of RDF typically have property
>objects that are subclass of resources - is that still an accurate
>design?

Yes.

>3. I've seen no last call comment that justifies the change
>
>The claim in the change log that it does not affect entailments is
>false.

The only document in the entire RDF document suite which mentions 
simple entailment is the semantics doc, and all the lemmas in section 
2 still hold.

>Test case:
>
>   sss ppp ooo .
>
>rdf entail
>
>   ppp rdf:type rdf:Resource .
>
>I believe the answer should be yes, but in any case the answer is
>distinguishable in RDF.

The answer is yes.  No RDF entailments are affected by this change.

If you really feel that this is a serious matter then I can go back 
and undo this, but I would rather not, as this makes it clearer that 
it is RDF itself which imposes this condition on IP (by virtue of the 
'type Property' condition on properties) rather than the graph syntax 
in some mysterious way.  There is no need for this condition to be 
imposed on simple interpretations, is the point -  it is only needed 
for RDF itself - and removing it from there makes the RDF conditions 
less redundant and  makes simple interpretations more like 
conventional first-order interpretations.

This is really all a part of the general organizational clean-up made 
since last call where conditions that apply purely to the RDF 
vocabulary are all labelled as RDF conditions, for RDFS are labelled 
RDFS conditions, and so on, ie it distributes content into different 
parts of the document. I really do consider it a technicality within 
scope for an editorial change.

Pat

-- 
---------------------------------------------------------------------
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, 30 July 2003 15:18:16 UTC