Re: LANG: new version of abstract syntax/translation document

A version of the document with all the fixes mentioned below is available
at  http://www-db.research.bell-labs.com/user/pfps/owl/specification.html


From: herman.ter.horst@philips.com
Subject: Re: LANG: new version of abstract syntax/translation document
Date: Mon, 15 Jul 2002 18:10:19 +0200

> The main additions to the current version of the document
> (version of 6 July), compared to the version that I reviewed earlier, 
> are about OWL Lite and the mapping to triple syntax.
> Below you find my review comments for the current version.

> - Given the discussions, I assume that the title will read 
> "Abstract Syntax of the OWL Web Ontology Language", or something 
> close to that.

Done.  Title changed to 
	OWL Web Ontology Language 1.0 Abstract Syntax

> - Section 6 is named "Mapping to the Triple Syntax".  Which triple
> syntax?

This is a good question.  The mapping is to an n-triple syntax for
DAML+OIL.  I have changed it to an n-triple syntax for OWL,
compatible with the Reference Description.  

> As there is nothing between RDF and the ontology layer in the
> layering envisioned for the Semantic Web, one would expect that
> only the namespaces rdf, rdfs and owl are used, as in the
> OWL reference description document.
> However, instead of the namespace owl, heavy use is made of 
> properties and classes from the namespace daml.

This has been changed to the OWL namespace.  I did not have easy access to
the Reference Description when I made the document.

> A complete description of the mapping from OWL abstract syntax 
> to RDF and RDF Schema triples also requires the definition of 
> these daml entities, preferably renamed with owl.
> At the very least, this document requires a reference to
> another document for these daml entities.

A reference to the Reference Description should suffice, and has been added.

> - In the table specifying the translation to triples in Section 6, 
> the restrictions atleast, atmost, and exactly should be replaced 
> by mincardinality etc.

Done.

> - In Section 5.3.3, in the cardinality production, two lines,
> containing the second occurrence of mincardinality and of 
> maxcardinality, should be omitted.

No.  Without this third alternative there would be no way of providing, for example,
minimum cardinality 5 and maximum cardinality 7 in one restriction.

> - The next to last paragraph before Section 1.1 should be changed,
> as RDF triples are now included.

Done.

> - Unlike the earlier version of the document, the productions 
> now contain many semicolons, which I prefer to leave out
> again.

This appears to be an artifact of the processing that the mail message was
subject to.  Long lines were incorrectly wrapped at some stage and the
wrapping introduced the semicolons.  

> - The second to last paragraph before Section 5.1 says that
> "each class axiom contains a collection of more-general classes,
> a collections of ...restrictions, and a collection of descriptions".
> This was true of the version of the document that I first reviewed
> but not of the current setup of the OWL class axioms, nor of the 
> current OWL Lite class axioms.
> I suggest to reconsider the text before Section 5.1 so that it deals
> with (current) OWL and OWL Lite, and largely contains information 
> common to OWL and OWL Lite.

I have changed this section somewhat.  In particular, I revised the
paragraph you mention.

> - Typos: ragnes, vocablary

Fixed.

> Herman ter Horst

Thanks,

peter

Received on Monday, 15 July 2002 13:41:48 UTC