Quick comments on the QRG

If you are going to keep the pointers to NF&R, then a simple ? is all
that is needed.  The bracketing parentheses do not add anything (and
only detract).

Organization:
  I disagree with putting class/... axioms in the ... sections but not
  enough to agitate for a reorganization.

Nomenclature:
  If you are going to use short forms, you should use "standard" ones,
  i.e., (...) instead of [...] (even though [] is nicer).


Semantics:
  If you have "semantics" for some axioms, you should have them for
  all.  If you can't have them for all, you shouldn't have any.  If you
  have them for all, then you should get the semantics at least close to
  correct. 

Links:
  The links to the Primer are broken due to the ongoing rewrite of the
  Primer.  

Abstract (2nd paragraph):  
  Much better would be:
    This document provides a quick reference guide to the OWL 2 language.

S1
  The initial bit of S1 should be something like:
    The standard ... in OWL 2 are

S2
  The title should be something like:
    OWL 2 Constructs

  It is not necessary to repeat the section titles.

S2.1.1
  Should be titled something like:
    Boolean Connectives and Enumeration

S2.1.2
  It is not necessary to have
   Every owl:Restriction is an owl:Class.
  as this comes from the structure of the document.

  A better arrangement for the cardinalities would be to have two lines
  in the second column, the first without the C and the second with.
  This would allow the removal of the "Cardinality Restrictions" box.

  if C presents -> if C is present

  It would be better to have the if ... lines left-justified and the
  triples below them indented a bit.

S2.1.3 
  Many of the points for S2.1.2 apply here as well.

S2.1.4
  See "Organization" note.

S2.2
  The introductory paragraph can just be replaced with
    Built-in datatypes are unary data ranges.

  You need to say that the D in DatatypeRestriction is a built-in
  datatype, arbitrary data ranges are not allowed.

  The table has some glitches.  It would probably be better to not have
  the f/v box by itself in a column.  (I'm not sure where it would be
  best to put it - perhaps in the left-hand column.)

S2.3.1
  The owl:ObjectProperty does not add anything here.

  The table has some boxing glitches.

  Better than = owl:Thing x owl:Thing is "Universal relation"
  Better than "empty binary relation" is "Empty relation"
  Even better would be to just remove the column.

S2.3.2
  There is no "," in the FS for DisjointObjectProperties.  Also occurs
  elsewhere. 

S2.4
  Many of the points for S2.3 apply here as well.

S2.5
  The j= doesn't need to be on a separate line.

S2.6
  This should not have the same status as, e.g., Declarations.  

S2.8 - S2.9
  This is not a good way of presenting annotations.  The problem is how
  to present annotations in the organization of the QRG.  The following
  appears to be the best compromise (but see "Organization" above).

  S2.8 Annotations

  S2.8.1 Annotations of Objects

  AnnotationAssertion( AP AS AV )
    AP AS AV

  S2.8.2 Annotations of Axioms

  AXIOM(Annotation(AP AV) ....)	  
    s p o .
    x rdf:type owl:Axiom .
    x owl:subject s .
    x owl:predicate p .
    x owl:object o.
    x AP AV .
      If AXIOM(...) becomes s p o . 

  AXIOM(Annotation(AP AV) ....)	  
    x ....
    x AP AV .
       If AXIOM(...) becomes x .... 

  S2.8.3 AnnotationProperties

  ....

  S2.8.4 Annotation Axioms

  .... (but without the "or" section)
   

S2.10
  As this is deprecated, it doesn't belong in this document.


S2.11

  I suggest instead

  S2.11 Annotations of Ontologies

    Ontology( ON [ VN ] Import(IN) ... Annotation(AP AV) ... ... )
     ON rdf:type owl:Ontology .
     [ ON owl:versionInfo VN . ]
     ON owl:imports IN .
     ...
     ON AP AV . 
     ...
     ...

     (Also for unnamed ontologies.)

S4.1
  owl:realPlus is gone
  owl:rational is in OWL 2
  need to discuss disjointness
  many of the time DTs listed are not in OWL 2


peter

Received on Tuesday, 17 March 2009 18:40:02 UTC