W3C home > Mailing lists > Public > www-webont-wg@w3.org > December 2002

Reference Review

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Tue, 31 Dec 2002 18:14:23 +0100
To: www-webont-wg@w3.org
Message-Id: <200212311814.23821.jjc@hpl.hp.com>


Version reviewed 1.99 2002/12/19 06:49:47

Scope of review mainly:

http://lists.w3.org/Archives/Public/www-webont-wg/2002Dec/0320.html


This document misses a clear definition of OWL Lite, OWL DL and OWL Full. It 
is inappropriate to point to the guide as given that definition since the 
guide is non-normative, and is unlikely to go into sufficient detail 
(currently it does not). The OWL Lite Features document is also inappropriate 
since it explicitly does not talk about syntax while OWL Lite is 
fundamentally a syntactic restricttion of OWL Full.

The relationship to RDF/XML is also insufficiently detailed.
The reader of this document would be unpleasantly surprised by an OWL DL 
compliant tool that refused to process some RDF/XML that uses a bnode as the 
object of two triples.

The discussion of Datatyping in this document is consistenly outdated. 

New sections needed
==================

Levels of OWL
Giving defintions for OWL Full, OWL DL and OWL Lite

Relationship with RDF/XML
+ Indicating that bnodes cannot be used freely in OWL DL and OWL Lite but that 
skolemization may be needed)
+ Indicating that in OWL DL and OWL Lite additional typing triples are 
required, and that, for instance, a range triple does not imply that its 
object is an owl:Class but this must be explicitly stated.

Detailed comments
===============

1. OWL Lite, OWL DL, OWL Full bullet point - neither of the references given 
are adequate. This document needs to expand this bullet point itself.

*Mising OWL with arbitrary RDF*

Misleading.
1: OWL Full does assign meaning to the additional triples.
2: OWL DL permits additional triples which do not "instantiate the schema 
defined" (I am not aware of any definition of *instantiate* w.r.t an RDF 
schema).
3: The additional triples allowed in OWL DL and OWL Lite are by no means 
arbitrary. (See my discssion of Dublin Core in my Feature Synopsis Review).

Stylistic comment
I am adverse to links from a recommendation off to an issue list.

2.

Last call documents should not have editorial to dos left in them.

"An OWL ontology consists of ..."
this sentence is misleading since OWL DL and OWL Lite ontologies are subject 
to many more syntactic restrictions. Also suggest delete the word followed, 
since an RDF graph is undered.
e.g.
"zero or more headers, class elements, property elements and instances.
OWL Lite and OWL DL impose further syntactic constraints."

(A discussion of these constraints is needed in this document).

Version Information - out of date 

Imports 
--------
needs discussion of how it works with OWL DL and OWL Lite.
Can an OWL Lite document import an OWL Full document?
Needs to point out that the class and property declarations required by OWL DL 
and OWL Lite cannot be imported but must be explicitly stated within the OWL 
DL or OWL Lite document.


Objects and Datatype Values
---------------------------------
"Datatypes are not OWL individual objects"
They are in OWL Full.

Enumerations
---------------
Missing an example corresponding to the abstract syntax 
<dataRange> ::= oneOf({<typedDataLiteral>} )

Property Restrictions
-----------------------
Suggest delete "currently" inappropriate in a recommendation.

Property elements
--------------------
The first paragraph could make it clearer that OWL Lite and OWL DL require 
that the property element is declated as of type owl:ObjectProperty or 
owl:DatatypeProperty.



Instances
-----------
Deeply misleading,
without discussion of the additional restrictions imposed by OWL DL and OWL 
Lite.

Datatype values
------------------
Incorrect, and needs substantial revision.


Appendix C
why is this a .txt document and not a .rdf document?


Jeremy
Received on Tuesday, 31 December 2002 12:16:56 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:56 GMT