- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Mon, 15 Jul 2002 10:13:27 -0400
- To: dlm@ksl.stanford.edu
- Cc: www-webont-wg@w3.org
Commentary on
Feature Synopsis for OWL Lite and OWL
Version of 8 July 2002
Problems:
- The abstract contains the same sort of inflated claims that pepper the
RDF documents. It should be changed so as to not implicitly claim that
OWL can handle arbitrary logical content. Similar problems occur in the
first paragraph of the introduction, which repeats the first 2 1/2
sentence of the abstract verbatim, a definition no-no. The abstract is
too long in any case.
- The introduction is rather repetitive. In particular, it introduces OWL
Lite twice. Among other changes, the last paragraph could be removed.
- The numerous design goals for OWL Lite are contradictory, and do not
correspond with the existing language. It would probably be better to
drop the extension design goal.
- The listings of feature constructors need more introduction, i.e., more
is needed at the beginning of Section 2.1. (What is a ``feature
constructor'' by the way?)
- The feature constructors do not correspond to those in the reference
description document, nor do they correspond to those in the abstract
syntax document.
- The document assumes a rather high level of understanding of its readers,
for example
- in its use of ``description'' and related terms near the beginning of
Section 3
- in its use of ``model'' near the beginning of Section 3.1
- in its use of ``sentence'', e.g., in Section 3.6
This is fine in a reference or formal document, but not very good in
a synopsis document.
- The document talks about ``valid'' OWL statements. What does this mean?
- The Mammal example is very confusing. What is the relationship between
``Mammal'' and ``THING''? What is ``the description THING''?
- What is the relationship between the tokens in Section 2 and the headings
in Section 3? What is the relationship between these and anything in OWL
Lite or OWL? This points out a very serious problem with the document -
its purpose is not given. Is the document trying to provide a high level
description of the features in OWL Lite and OWL? If so, what are the
comments about abstract syntax? Further, the high-level descriptions do
not correspond to either the abstract syntax document or do the reference
description document. (See, for example, the Class and Property entries
in Section 3.1.) Is the document trying to describe the abstract syntax
or the triple syntax? If so, where are these relationships made clear?
This confusion is made clear in Section 5, where the first sentence talks
about a feature synopsis (not syntax) but the second talks about
constructors (syntax).
I recommend that this document use precise names (like functional) for
all these things, and try hard not to use the DAML+OIL or OWL names. The
document would be limited to providing a very high level description of
the functionality provided in OWL Lite and OWL, and would not mention any
syntax at all. As part of this, Section 2 would be removed. Discussion
of functionality would not be organized around syntax (e.g.,
minCardinality, maxCardinality, and cardinality in Section 3.4) but would
instead be organized around functionality (e.g., Section 3.4 would just
talk about cardinality, with min- and max- being mentioned inline). Two
examples of how this could work are Section 3.5 (without the last
sentence and the gratuitous one-element list) and the paragraph on oneOf
(without the heading).
- The introductory sentence in Section 3.3 does not correspond to the first
element of the list. Similarly, the last two entries on this list have a
very different behaviour from the previous four and deserve to have a
separate introductory bit.
- The document is rather confused about classes and descriptions. For
example, classes are used in Section 3 in contrast to descriptions, but
in Section 4, classes appear to include descriptions. In Section 4, the
phrase ``complex class descriptions'' is used as well.
Minor problems:
- The title should probably be changed to be more in line with the reference
description.
- The formal spec document does not contain a formal semantics for OWL.
- The constructors should probably have owl: in front of them, where
appropriate, otherwise, rdf: or rdfs:.
- functional is used instead of unique
- The comment on hasValue is misleading. There is no hasValue at all in
OWL Lite.
- The sentence about instance probably does not belong in the paragraph
about Property.
- This will be the first version of the document, and thus there should not
be a history.
Editorial problems:
- ``a languages''
- The paragraph starting ``This smaller language'' should instead start
``OWL Lite''.
- The first paragraph of Section 2 does not read well. The italicized
terms term should probably be italicized.
- The individual feature is not italicized. Is this correct?
- Section 3 exists, so it discusses something. Stating that it ``will
discuss'' something gives the impression that it does not now do so.
Similarly, ``proposed'' gives a very tentative impression.
- The first sentence of Section 3 should reference OWL Lite, as in the
``language features of OWL Lite''.
Peter F. Patel-Schneider
Bell Labs Research
Received on Monday, 15 July 2002 10:13:37 UTC