RE: how the guide treats classes as instances (5.19)

Dan,

Good points.  Some are harder than others.  With input from you and
Guus I would suggest the following.

1. When we first mention the wine ontology we should say that it
satisfies the requiremens of OWL DL.  Some examples will focus
on the use of OWL Full capabilities, and they will be marked as OWL Full.

2. Then, when we hit 

  The wine ontology as it currently exists
  would require the ability to treat classes
  as instances

We can explain that while wine.owl satisfies OWL DL, it could be
extended into OWL Full to provide this support.

3. Eliminate the "will be of interest to the advanced user," language.

4. "Er... InverseFunctional Datatype properties are as common as
falling off a log; they're called database keys in other contexts."

Need to include an explanation of this.

- Mike

-----Original Message-----
From: Dan Connolly [mailto:connolly@w3.org]
Sent: Thursday, November 14, 2002 9:43 AM
To: www-webont-wg@w3.org
Subject: how the guide treats classes as instances (5.19)



The one thing remaining before I'm happy to
close issue 5.19 is a good treatment in the
guide.

This is what I was looking for as a result of...

> ACTION: Guus, Frank: to move the issue forward, will write up 1-d
> and 2-d views to make clearer to users.

What they wrote 31 Oct is clear enough to the WG...
http://lists.w3.org/Archives/Public/www-webont-wg/2002Oct/0310.html
but I don't think it's text we expect users
to understand.

While the Guide does a great job on many issues,
I disagree with Frank about the treatment
of the species of OWL...

 "In general, I think it is fine"
 -- Frank van Harmelen, 7 Nov
http://lists.w3.org/Archives/Public/www-webont-wg/2002Nov/0062.html

Text like

  The wine ontology as it currently exists
  would require the ability to treat classes
  as instances
http://www.w3.org/TR/2002/WD-owl-guide-20021104/#Introduction

suggests you can't express classes as instances
in OWL; but you can; You just don't get predicable
reasoning time. The fact that you can do classes
as instances is in the document, but not until
later; and it treats it as some obscure that
"will be of interest to the advanced user,"
while WG discussions have made it clear that
this comes up routinely, and users need
to know about it.

That is, I don't think this text reflects
our discussions:

  Another significant difference from OWL DL is that
  a DatatypeProperty can be marked as an
  InverseFunctionalProperty. These are differences that
  will be of interest to the advanced user. This document
  does not describe the use of these features.

Er... InverseFunctional Datatype properties are
as common as falling off a log; they're called
database keys in other contexts.

With apologies for not providing specific text
(I haven't found time, but I see this issue
is on our agenda today), I ask that the
guide treat OWL DL and OWL Full as peers,
especially w.r.t. classes as instances.

One approach would be to show the more
straightforward expression of the wine ontology
using classes as instances, then note
the unpredictable nature of reasoning
using those idioms, and then show how
to re-do it within the OWL DL constraints.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 14 November 2002 13:30:15 UTC