Re: pfps-12 lists are not well formed

From: Brian McBride <bwm@hplb.hpl.hp.com>
Subject: pfps-12 lists are not well formed
Date: Sat, 24 May 2003 13:58:39 +0100

> Peter,
> 
> Danbri and I have been discussing how to resolve your issue about the 
> wellformedness of lists:
> 
>    http://www.w3.org/2001/sw/RDFCore/20030123-issues/#pfps-12
> 
> We are proposing to add the following note to the text at:
> 
>    http://www.w3.org/TR/rdf-schema/#ch_collectionvocab
> 
> as the last paragraph.
> 
> [[
> NOTE:  It is possible to construct RDF graphs that use the RDF collections 
> vocabulary to partially describe a list.  Similarly there are graphs that 
> use this vocabulary in a way that is consistent with the RDF(S) formal 
> semantics, yet do not represent "well formed" lists.
> ]]
> 
> We considered trying to trying to provide a full prose account of the 
> wellformedness of lists, but are currently disinclined to attempt such an 
> intricate task in natural language.
> 
> Will adding this note address your concern.  If not, could you please 
> suggest alternative text that you would find more satisfactory.
> 
> Brian

I fail to see how this response addresses my comment.  

I don't see how it addresses

> The RDF Schema document provides intended meanings for some of the RDFS
> vocabulary that is not supported by the RDF Semantics.  Vocabulary that
> fits into this category includes rdfs:label and rdfs:comment.
[from http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0126.html]

I don't see how it addresses

> Consider the following three examples (slightly reformatted but otherwise
> unchanged):
> 
> 	rdf:type is an instance of rdf:Property that is used to state that
> 	a resource is an instance of a class.  A triple of the form:
> 		R rdf:type C
> 	states that C is an instance of rdfs:Class and R is an instance of C.
> 
> 	rdf:first is an instance of rdf:Property that is used to indicate
> 	the first item of a list.  A triple of the form:
> 		L rdf:first O
> 	states that L is an instance of rdf:List and that O is the first
> 	item of the list. 
> 
> 	rdfs:label is an instance of rdf:Property that is used to provide a
> 	human-readable version of a resource's name.  A triple of the form:
> 		R rdfs:label L
> 	states that L is a human readable label for R.
> 
> There is essentially no difference between the way these three are worded.
> However, the first (rdf:type) is a fundamental part of the semantics of
> RDF.  There are semantic conditions in RDF that make the description above
> for rdf:type part of the very meaning of RDF.  The second (rdf:first) and
> third (rdfs:label), on the other hand, have a very different status.  There
> are no semantic conditions that force the descriptions above for these two
> vocabulary elements to play the roles given for them.
[from http://lists.w3.org/Archives/Public/www-rdf-comments/2003JanMar/0133.html]

The point of my comments here has always been that there are parts of the
RDF Schema document that go beyond what is supported by the RDF semantics.
I believe that these parts of the document should be changed, and that
changes to other parts of the document will not suffice to override these
over-reaching parts of the document.

For the case of rdf:first above, I would much prefer

 	rdf:first is an instance of rdf:Property that can be used to build
 	descriptions of lists and other list-like structures.  A triple of
 	the form:
 		L rdf:first O
 	states that there is a first-element relationship between L and O.

	Note:  RDFS does not require that there be only one first element
	of a list-like structure, or even that a list-like structure have a
	first element.

I note that similar changes would have to be make for at least rdf:rest and
rdf:List.


Peter F. Patel-Schneider
Bell Labs Research

Received on Sunday, 25 May 2003 07:26:11 UTC