Re: comments on handling comments

On Thu, 2003-04-03 at 10:58, Peter F. Patel-Schneider wrote:
> I'm including exerpts from a recent exchange on the comments mailing list
> to illustrate some issues I see in handling comments.
> [...]
> > 1 - OWL is an extension of RDF(S) -- that is, all RDF and RDFS 
> > documents are legal OWL Full documents and all OWL documents are 
> > legal RDFS documents.  However, OWL extends the vocabulary of RDFS to 
> > allow some more expressivity.  For example, in OWL you can say that a 
> > property is required (owl:minCardinality of 1) or optional 
> > (owl:minCardinality of 0) and other such things.
> This is only a partial answer, and resulted in further queries.

Indeed, JimH got a rather urgent and pointed reply from me for
doing that. (Meanwhile, it could have been worse; if JimH
just hadn't done anything, he wouldn't have learned anything
and the commentor might be wondering... 'hello? anybody home?')

DON'T re-write the spec in answer to comments.

Just point the commentor to the relevant part of the spec.
After taking the pointy stick out of JimH's side, I
tried to set an example...

excerpt from
Re: OWL Questions! (RDFS/OWL relationship) Dan Connolly (Wed, Apr 02 2003)

> 1) OWL is an extension of RDFS? That is, OWL has much expressive power
> than RDFS ( i can express thing that with RDFS i coudn't ) or it's
> equal but it can express relation and properties more easily? 

The overview covers this question...

"OWL has more facilities for expressing meaning and semantics than XML,
RDF, and RDF-S, and thus OWL goes beyond these languages in its ability
to represent machine readable content on the Web."

  -- 1. Introduction
  OWL Web Ontology Language 
  W3C Working Draft 31 March 2003

and the reference covers it in a it more detail...

"OWL is a vocabulary extension of RDF."
  -- 1.4 OWL and RDF semantics
  OWL Web Ontology Language 
  W3C Working Draft 31 March 2003

The exact nature of the relationship is covered in
excruciating detail in the semantics document,
and there are a number of relevant test cases.

But I suspect the two sections above are most responsive
to your question. Do they answer your question well enough?

> In my opinion answers in the comments mailing list should be as complete as
> possible. 

For now, don't try to answer using words other than the ones already
in the spec.

We're trying to confirm that our specs are good enough *as is*.

Often people just haven't managed to find/read the part
of the spec that's most relevant to their comment.

> >    That said, OWL does have some special subsets that are identified 
> > in our documents (OWL Lite, OWL DL and OWL Full) - two of which have 
> > some special restrictions, but some nice properties for reasoning 
> > systems. Not all RDFS documents are necesssarily in OWL Lite or OWL 
> > Full.  Details can be foudn in our documents.
> This is incorrect.  Only really a typo to be sure, namely using OWL Full or
> OWL DL, but this still caused followup.

Right; the last thing we want to do is spend time debugging
text that isn't even part of our spec.

> [...]
> My suggestion is that there should be no response in the comments mailing
> list without some review of the response.

You're welcome to get review in all cases.

Editors and the chairs are also allowed, at this point, to reply
ala "does this part of the spec answer your question?".

>   The only exceptions I can see
> would be 
> 1/ formulaic responses thanking the commenter
> 2/ responses pointing the commenter to a separate informal response
>    elsewhere
> peter
Dan Connolly, W3C

Received on Thursday, 3 April 2003 13:27:34 UTC