Re: [closed] pfps-05

From: pat hayes <phayes@ai.uwf.edu>
Subject: Re: [closed] pfps-05
Date: Sun, 1 Jun 2003 22:39:24 -0500

> >From: pat hayes <phayes@ai.uwf.edu>
> >Subject: Re: [closed] pfps-05
> >Date: Fri, 30 May 2003 17:20:17 -0500
> >
> >[...]
> >
> >>  What would it take to convince you? How many entailments do you want
> >>  to be convinced of? Your requirements for being satisfied seem
> >>  open-ended and expandable. Unless I have some indication what you
> >>  will accept as a sufficient condition to satisfy this comment, I do
> >>  not propose to set out to offer any further response.
> >>
> >>  Pat
> >
> >Well, then don't expect any sign-off from me.
> >
> >What do you want, a blank cheque?  The current document is incomplete and
> >internally inconsistent.
> 
> I do not want a blank cheque, as you put it, but we are obliged by 
> W3C protocols to deal with comments in isolation to some extent, so 
> if you are unable to sign off on any comment until you are satisfied 
> on everything, then we probably never will reach sign-off by you. 
> Brian is anxious for us to get the issues dealt with, and we seem to 
> have to do this by a piecemeal process. I was not asking you to 
> approve the entire document, only the part of it which addressed your 
> comment.

My comment was that the RDFS closure rules are incomplete.  There is no way
that I can sign off on this until I see a finished version of the model
theory and the closure rules, which I have not yet seen.  From my end it
looks exactly like you are asking for a blank cheque.

If the process is getting in the way of doing the right thing, then I
suggest that you get the process changed.

> Le me suggest that we compromise. The only parts of the document 
> which affect OWL are the normative parts.  None of the rules tables 
> or appendices are normative. (I note that if the OWL documents were 
> held to the same standard that you are holding me to, then AS&S ought 
> to include complete and correct systems of inference rules for all 
> three versions of OWL together with completeness proofs. ) 

This is a false and misleading statement.  If S&AS contained inference
rules that it claimed to be complete and correct, then yes, they had better
be complete and correct and there had better be something to back up the
claim.  However, S&AS has no such inference rules.

This, of course, gives rise to a simple solution to the ongoing problems
with the closure rules - just remove them.

> So let me 
> suggest that we agree that you will accept a piecemeal approach to 
> settling issues which arise from parts of the document which are 
> informative, and I will agree to a somewhat higher standards of 
> finish before asking you to approve any repairs done to the normative 
> parts of the document. Would that be acceptable?

This would not be acceptable.  There is no way that I can sign off on the
correctness of the RDFS closure rules until I see a finished version of the
model theory.  

> Pat

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Monday, 2 June 2003 10:41:53 UTC