Re: draft responses for four JC LC comments

From: Sandro Hawke <sandro@w3.org>
Subject: Re: draft responses for four JC LC comments 
Date: Wed, 13 May 2009 00:42:18 -0400

>> I've put together draft responses for four of the JC LC comments.
>> 
>> See http://www.w3.org/2007/OWL/wiki/Responses_to_Last_Call_Comments for
>> more information.
> 
> A good start.   Some thoughts...
> 
> re JC1 - I had to go back and re-read Jeremy's comment; I had missed
> that he mentioned the [name], where you only talk about the
       	  	    	^^^^
> namespace.  I guess I'd add something like: "The Working Group believes
> this new language is the successor to OWL 1, and as such ought to use
> the OWL name."

I think that if this is said it should be said even stronger.

  The Working Group has produced OWL 2 as the current version of OWL,
  and as such it ought to use the OWL name.

I've adjusted the response.


> re JC2 - very nice

Probably JC3.  :-)

> re JC4 - Jeremy refers to those documents already.  I guess I'd ask him
> in what way these two sections...
> 
>     http://www.w3.org/2007/OWL/wiki/New_Features_and_Rationale#Features_.26_Rationale
> 
> and
> 
>     http://www.w3.org/2007/OWL/wiki/Quick_Reference_Guide#New_Features_in_OWL_2
> 
> fall short of what he wants, and how (specifically) he'd like them
> changed.  (I don't think there's a New Features bit in Primer any more.)

We could point out the two sections, perhaps, as places where new
features are directly called out, but the Primer does mention the new
features of OWL 2 in a way accessible to "customers", albeit it does not
call them out as new.

> re JC5 - I'm trying to understand why the range of owl:predicate needs
> to be unrestricted.  If it were rdf:Statement, then, yes, there'd be an
> entailed triple, but couldn't that triple actually be required/suggested
> in the input.  If this is explained in some e-mail or minutes somewhere,
> or even the spec, it should be linked from JC5.  (I expect this matter
> will be an OWL 2 FAQ, because RDF reification has great associated
> drama.  I'd love a nice written answer, if we do need to keep this.  If
> the domain/range of RDF reification is really just broken [cf your
> message tonight about it], we can fix that using the errata process
> rather than shun RDF reification.)

I don't know why the range of owl:predicate *must* be unrestricted.  A
partial reason is to cut down on the inferences that can be made from an
annotation.   I don't think, however, that this is central to the
response.  It is sufficient to point to the difference with respect to
the domain of owl:subject, owl:predicate, and owl:object.

There is nothing technically wrong with the domain/range of RDF
reification predicates, just that they are different than the ones for
owl:subject, owl:predicate, and owl:object.  The problem in the RDF
documents is that one of them informally says that there is an inference
to be made on objects of rdf:predicate triples, but that this is not
picked up in the range of rdf:predicate nor as a special-purpose
inference in the RDF semantics.  This is simply an RDF-internal
inconsistency between the explanatory text and the formal treatment.  It
would be easiest to deal with by changing the explanatory text, of
course, and that would be consistent with the documents' history and
change logs, but that somehow feels less-than-ideal to me.

>      - Sandro

peter

Received on Wednesday, 13 May 2009 10:10:01 UTC