# Re: Review of Sean Bechofer's OWL Doc

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Wed, 17 Dec 2003 23:32:46 +0100

Message-Id: <200312172332.46476.jjc@hpl.hp.com>


I finally got round to this ...

Summary: a great document - definitely fit for purpose.

My review consists of minor comments and corrections.

===

doc reviewed

http://wonderweb.semanticweb.org/owl/parsing-03-10-09.shtml

a scribbled on printout can be found at

http://lists.w3.org/Archives/Public/www-archive/2003Dec/0036

(as a pdf)

Bits that might be most unclear:

Page 1:

(A) this is suggestion to move text and intro of scope of document from page 2

page 2 errors
I prefer the following text or similar for internal and external error
definition
[[
There are, in general, two ways in which an RDF graph may fail to
correspond to an OWL Lite or DL ontology.

\begin{itemize*}
\item There is an OWL Lite or DL ontology in abstract syntax form
which maps to a superset of the given triples
but some of the triples
have been forgotten and are not in the graph.
\item The ontologies in abstract syntax form that map to the
triples or any superset of the triples
violate some of the restrictions for
membership of the OWL Lite or DL subspecies. (This includes
the case where there are no such ontologies).
\end{itemize*}

We might (loosely) describe the first as \textit{external} errors, and
the second as \textit{internal} errors.
]]

(possibly familiar ...)

Bottom of page 3 - you have already excluded this para from scope, it is the
thin end of a thick wedge

General comment - the terms used in the abstract syntax are not identical to
those in S&AS - probably does not matter but maybe needs a disclaimer
somewhere

Page 6 the note is to me about possible bug in my checker

(looking at S&AS and comparing I note that DeprecatedClass is also permitted
but I don't think you treat those at all in this paper, a plausible
omission).

Page 7 it might be worth at some point in this paper echoing the transitive
property constraint from S&AS

Page 7 bottom - I think it is helpful to specify when individuals must be
urirefs and when they can be unnamed (blank) this is not obvious (IMO).

Page 10 - rdf:nil does not have any triples at all - it is just a special
uriref

Page 13 illegible scribble is "rdfs:Literal" and "n is blank"

Page 15 top - you have omitted the two cases blank node as object of
rdfs:subClassOf and blank node as subject of rdfs:subClassOf

(also a *h*as)

Page 16 - the disjoint stuff is OK (except for the glaring typo with
"EquivalentClasses"). Note the very last condifion is unnecessary because c
and d must be urirefs by the operation of the algorithm.

scribble on page 17 is
"or datatype, ontology or unnamed ontology or individual"
(I wonder whether it might be simpler to say what cannot be annotated - hmmm
probably not)

Jeremy

Received on Wednesday, 17 December 2003 17:36:28 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:03 GMT