# Review Direct Semantics

From: Thomas Schneider <schneidt@cs.man.ac.uk>
Date: Sun, 5 Apr 2009 15:40:12 +0100
Message-Id: <02AB4970-638F-4445-822B-0B854583050E@cs.man.ac.uk>
To: OWL Working Group WG <public-owl-wg@w3.org>
Hi Boris, Peter and Bernardo,

Cheers

Thomas

============================================================
Review for "OWL 2 Web Ontology Language -- Direct Semantics"
============================================================
revision 21904
http://www.w3.org/2007/OWL/wiki/index.php?title=Semantics&oldid=21904

Summary
-------

I find this document clear, precise and technically correct, and have

-----------------

* 1 Introduction

- 2nd sentence: This is much clearer than before. However, I have
stumbled over "is compatible with", which is quite lax and not a
standard term in this setting.  It is of course explained in the
following sentences, but why not say "extends" in the first place?

- 4th paragraph, 1st sentence: It might be clearer to repeat "of",
i.e., "annotations of ontologies, *of* anonymous individuals, *of*
axioms, and *of* other annotations". Otherwise an unexperienced
reader might misunderstand the meaning of the sentence as "OWL 2
allows for annotations of ontologies, *for* anonymous individuals,
*for* ...".

* 2.1 Vocabulary

- Here and in the following subsections, I'm still not happy with
"(C)^C" and "(DT)^{DT}". I know we have discussed quite a few
alternatives and ruled them out for several reasons. But still I'm
sure that "(C)^C" etc. will confuse non-experts, although I don't
have a suggestion for a new solution.

* 2.2.2 Data Ranges

- The remark that data ranges can be n-ary doesn't make it clear
that this document (following the Specification) only considers
the unary case. For a moment, I was expecting to see the
interpretation of an "atomic" n-ary data range. The wording in the
Specification [ http://www.w3.org/2007/OWL/wiki/Syntax#Data_Ranges
, 4th sentence ] makes this clearer than the wording in the Direct
Semantics; perhaps it should be copied.

* 2.3 Satisfaction in an Interpretation

- What is the purpose of the last occurrence of "the axiom" in the
1st sentence? If "axiom" has to occur three times in this
sentence, it might be given a name. I remember that we had issues
with greek letters not being rendered correctly, but why not call
the axiom A, Ax, or ax?

* 2.3.5 Keys

- Why not replace "keys" (in heading, sentence and table caption)
with "key axioms"? This would be consistent with the 2nd sentence
of 2.3.

* 2.5 Inference Problems

- Now that all inference problems are defined w.r.t. a datatype D,
we have the problem that the term "model" is not defined w.r.t. D.

- 3rd last sentence: I've added a reference to the definition of a
simple OPE to the word "simple".

* Throughout

- When viewing the document with sans-serif fonts, the capital
letter I and the digit 1 can hardly be distinguished. This
complicates reading some of the expressions used in the document,
e.g. those in the 3rd and 4th line of Table 4. In 2.4, you even
use I_1, which contains \Delta_I and \cdot^{I_1}, but all
subscripts read like the capital I.

I remember that Boris has mentioned this problem in the last
discussion, but I don't think we've fully discussed the
alternatives here. I'm aware that possible changes can be
far-reaching and therefore require a lot of work, but still I'd
prefer to avoid confusion whenever possible. So how about using
the capital "J" instead of "I"?

+----------------------------------------------------------------------+
|  Dr Thomas Schneider                         schneider@cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+