Re: Proposed Consensus Review of RDF Core Documents

From: Jim Hendler <hendler@cs.umd.edu>
Subject: Proposed Consensus Review of RDF Core Documents
Date: Thu, 20 Feb 2003 06:20:08 -0500

> 
> Guus and I have worked (with advice from Dan) to boil down our 
> reviews of the RDF documents into a single proposed consensus review. 
> We  will discuss on the telecon today (Feb 20).   We will ask Herman, 
> Raphael and Jean-Francois to post their individual reviews directly 
> to www-rdf-comments@w3.org, with this summary being our consensus 
> comments.
> 
> 
> 
> ================
> RDFCore LC documents
> Response by the Web Ontology Working Group (draft)
> 21 Feb 2003
> 
> --------------
> RDF Design Issues
> --------------
> 
> i. Design of rdf:XMLLiteral and rdf:parseType="Literal":
> 
> The full integration of this feature of RDF into OWL necessitates that the
> denotation in the domain of discourse be fully defined by the source RDF/XML
> file. We therefore request that you remove sufficient implementation 
> variability to ensure that this is the case.  An example fix would be 
> to require an RDF/XML parser to use a specific canonicalization on 
> input.
> 
> We have resolved that rdf:XMLLiteral will be a built-in datatype in 
> OWL Lite, OWL DL and OWL Full contingent on a satisfactory response 
> from RDF Core WG on this comment.

I don't remember this resolution.

> ii.Constraints on rdf:parsetype="Collection"
> 
> We would prefer that rdf:parsetype="Collection" would be allowed to 
> be a list of literals, not just a list of RDF node elements.  This, 
> would permit some constructs in OWL that are difficult under the 
> current design.
> 
> -------------------------------------------
> Consensus comments on the RDF Concepts Document
> --------------------------------------------
> 
> We believe the RDF Concepts Document is a useful document and helpful 
> in understanding RDF and its use.  However, out Working Group did 
> have some concerns with respect to the issue of social meaning as 
> discussed in this document.
> 
> The Web Ontology WG has mixed views on this issue and could not agree 
> on a specific consensus response in the time available.  However, we 
> note that a number of participants in the Webont WG have reservations 

Add ``strong'', perhaps?

> about the RDF view on the social meaning of RDF. For example, it was 
> felt by some to be unacceptable that two classes that differ only in 
> their rdfs:comment content would not entail each other.

This issue with rdfs:comment is completely different.  rdfs:comment does
impinge strongly on social meaning, but the issue of non-entailment here is
about formal meaning, not social meaning.

> We did reach consensus to request that the wording in the RDF Schema 
> and the RDF Concepts documents be rephrased to explain this issue, 
> and particularly its impact, more clearly, as this has ramifications 
> on other languages, such as OWL, which are extensions to RDF.

Which issue?  There are two here.

> -------------------------------------------
> Consensus comments on the RDF Schema Document
> --------------------------------------------
> We believe that the design of the language, as reflected in the LC 
> documents, is such that OWL can appropriately use RDF Schema and 
> endorse this design.
> 
> Raphael Volz of our group has prepared a detailed review of this 
> document which he will send to the RDF Core WG.  The Web Ontology 
> Working Group agrees with the spirit of his review and summarizes our 
> comments below:
> 
> i. Although this document is called RDF Schema we think that the 
> title "RDF Vocabulary Description Language" would be clearer, and 
> make the difference from XML Schema (used for validation) more 
> evident.
> 
> ii. The current design does not specify what the behavior is for 
> domain/range constraints stated on super-properties wrt. to 
> subproperties.  We would request that a default behavior be specified.

This paragraph does not address the issue that Raphael brought up.  The
interaction between domain/range constraints and sub-/super-properties is
well specified in the formal semantics, and not a separate problem.  The
problem has to do with Section 4 of the RDF Schema document, which vaguely
talks about RDF applications using domain/range constraints for things like
document validity checking.

> -------------------------------------------
> Consensus comments on the RDF Semantics document
> --------------------------------------------
> We believe that the design of the semantics, as reflected in the LC 
> documents, is such that OWL will be able to layer appropriately. 

The basic design of the semantics may be suitable, but there are many
problems in the details that affect OWL.  It might be able to layer OWL on
the semantics as described in the RDF Semantics last call working draft,
but it would require considerable work on our part to get around the errors
in that document.

> However, we have a number of concerns that need to be addressed to 
> improve the document (and, in particular, to fix some inconsistencies 
> in the current document).
> 
> Herman ter Horst of our group has prepared a detailed review of this 
> document itemizing inconsistencies he has found.  The Web Ontology WG 
> endorses the spirit of his review, and has asked Herman to help 
> insure that the final RDF Semantics document is edited to fix the 
> inconsistencies and editorial issues that he identifies.

You should mention here the many errors that I have found in this document,
most of which have been verified by Pat.  We might want to discuss my
current view of the way forward with respect to the RDF Semantics document
at the teleconference today.

peter

Received on Thursday, 20 February 2003 06:34:42 UTC