Review of Test Document

Guus and Jim asked me earlier this week to review the test
document. Due to time constraints, this is necessarily brief. My
apologies for that.


Review of Web Ontology Language (OWL) Test Cases.

The document is in good shape, and I have no major comments or criticisms.
A number of minor points are raised below.

On a general note, with the inclusion of all the test contents, the
document is now very large (>400 pages). Is the inclusion of test
contents absolutely necessary? I would support Jeremy's proposal of
[1] to exclude the test contents from the document.

Minor Typo

Section 6

	A test for the incorrect us of OWL namespace

Conformance and Testing Implementations

Section 4.2.2 states:

An OWL consistency checker SHOULD report network errors occurring
during the computation of the imports closure.


Note: Complete OWL DL consistency checkers and Complete OWL Lite
consistency checkers MAY return Unknown on an OWL DL document or OWL
Lite document in the case where a resource limit has been exceeded.

Section 5.2 last para says:

A complete OWL Lite consistency checker or a complete OWL DL
consistency checker should not return Unknown on the relevant
consistency or inconsistency tests.

I find it unclear from this what the expected behaviour of a complete
consistency checker should be in the light of network problems. For
example, the case where an OWL document is DL, but this can't be
determined due to an import problem, e.g. a resource is typed as an
owl:Class in an imported ontology, but the server serving the imported
ontology is inaccessible.

I suggest this can be clarified by either being explicit as to the
expected behaviour of an OWL Lite/DL complete consistency checker when
presented with an input which cannot be determined as being Lite/DL (as
this may only become apparent at the point when the check is being
performed), or extending 2. to explicitly cover network problems.

Approved/Proposed Tests

I have not checked the content of all the tests in detail, but have
identified some possible problems with one or two of them.

o allValuesFrom/conclusions001.rdf and
  allValuesFrom/nonconclusions002.rdf are claimed as OWL-Lite, but are
  not, as first:c is not typed.

o The sameClassAs and samePropertyAs tests use sameClassAs and
  samePropertyAs rather than equivalentClass and equivalentProperty.

  uses owl:hasClass. Admittedly the Manifest claims that this is a Full
  document, but I guess this is an error.

DL Tests

As the contributor of the DL tests, these are perhaps my fault(!),
but some of the DL tests in the approved section have broken
syntax. Several of them include things like:

 <owl:onProperty rdf:resource=""/>

This should rather be:

 <owl:onProperty rdf:resource=""/>
  <owl:Class name=""/>

or perhaps:

 <owl:onProperty rdf:resource=""/>
 <owl:someValuesFrom rdf:resource=""/>

In addition, If they are to be brought into line with B.4, the
"" domain should be replace with an example
domain as "" is not a retrievable web


Appendices A and B are neither marked as Informative or Normative. I
assume the former.

A.2 Approval

Many of the tests are proposed by Carroll and verified by De Roo's

Can you clarify what "De Roo's implementation" is here, and to what
extent the tests are verified. Alternatively, remove the statement, as it
doesn't really provide much information -- we already know that
Jeremy and Jos have done a lot of work on this document :-)

[The next point is a question perhaps relating to W3C policy rather than
the document.] What is the procedure (if any) for
extending/amending/maintaining the test suite once the language has been
accepted? Is the working group still in existence at that point?

<Section B.9

	The URL of the first file concatenated with #
	The URL of the second file concatenated with #

Please clarify - what are the first and second files in this context?




Sean Bechhofer

Received on Friday, 21 March 2003 07:07:48 UTC