W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > November 2003

RE: Test case document, simple entailment: preferred option.

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Mon, 10 Nov 2003 10:59:28 +0100
To: "Jan Grant" <Jan.Grant@bristol.ac.uk>, "RDFCore Working Group" <w3c-rdfcore-wg@w3.org>
Cc: <www-webont-wg@w3.org>
Message-ID: <BHEGLCKMOHGLGNOKPGHDAEFBCCAA.jjc@hpl.hp.com>


WebOntWG

there is a proposal to change some of the RDF test schema, which would have
some impact on the derived OWL testOntology.

Looking at

http://www.w3.org/TR/owl-test/#manifest

OWL tests use the rdfTest:entailmentRules property only indirectly.

Each OWL test has a test type, and most have definitions which restrict the
value of rdfTest:entailmentRules appropriately e.g.
[[
<rdfs:Class rdf:about="#PositiveEntailmentTest">
   <rdfs:subClassOf rdf:resource="&rdfTest;PositiveEntailmentTest"/>
   <rdfs:comment>
This is a positive entailment test according to the OWL entailment rules.
   </rdfs:comment>
   <rdfs:subClassOf>
     <owl:Restriction>
       <owl:onProperty rdf:resource="&rdfTest;entailmentRules"/>
       <owl:hasValue rdf:resource="http://www.w3.org/2002/07/owl#"/>
     </owl:Restriction>
   </rdfs:subClassOf>
</rdfs:Class>
]]

Thus, unless some developers actually do OWL processing of the manifest, to
access the appropriate entailment rules, we can make changes.

For example, the HP tests use the test class as an explicit indicator of
which testing software to run. Hence the implied values for
&rdfTest;entailmentRules don't actual enter into it.

Mr incredibly-commited-to-OWL might extract a test from the manifest, then
determine which rules to apply by taking the values for
&rdfTest;entailmentRules in the OWL closure of the manifest. I don't really
believe anybody does that! They would have to change their code.

In any case, this change would mean that there will be a knock on change on
the OWL testOntology.


Jeremy




> -----Original Message-----
> From: w3c-rdfcore-wg-request@w3.org
> [mailto:w3c-rdfcore-wg-request@w3.org]On Behalf Of Jan Grant
> Sent: 07 November 2003 18:01
> To: RDFCore Working Group
> Subject: Test case document, simple entailment: preferred option.
>
>
>
> In addition to the actions on this editor to withdraw a few test cases,
> Sandro has raised the issue of identifying simple entailment tests in
> the manifest.
>
> Currently these positive and negative simple entailment tests are
> distinguished by _omitting_ any test:entailmentRules entries on the test
> description.
>
> DanC, however, points out that the closed world assumption which is
> embodied in this choice makes running the test cases in some harnesses
> problematical.
>
> Although he didn't directly point it out, the current design uses that
> assumption when describing test cases using other entailment rules; in
> addition, such an assumption is required for dealing with the
> declaration of "supported" datatypes that a test case requires.
>
>
> A full "fix" for this would be to adopt the rdf Collection mechanism for
> describing the entailment rules and the datatype support of a test case.
>
> A prime motivation for the manifest format is that it should be readily
> usable without recourse to a full RDF parser (the RDF format is a
> "bonus"). At the time the test case manifest format was outlined, a
> convenient shorthand for collections wasn't present in the syntax.
> However, such syntax is now straightforward.
>
> The change to the syntax and appearance of a manifest file would be
> minimal. However, there are two large "cons" to this fix:
>
> - firstly, existing test case harnesses would need largish (compared to
> the alternative proposal) changes to accommodate this change
>
> - since the WebOnt test cases build upon the schema for RDF test cases,
> a change of this nature here would have an impact on the WebOnt test
> cases. (I think Jeremy expressed this better.)
>
>
> A simpler "fix" is the explicit adoption of a test:entailmentRules
> marker to test cases that utilise the simple entailment rules. In this
> case, the marker would be:
>
> <test:entailmentRules
> rdf:resource="http://www.w3.org/2000/10/rdf-tests/rdfcore/testSche
> ma#simleEntailment" />
>
>
> There are five test cases whose manifest entries would need to be
> amended with this markup. In addition, the testcase schema document
> would receive a description for this resource (which is a manifest
> constant indicating simple entailment rules should be applied). Finally,
> the following text would be added to the test case document where the
> Positive Entailment Tests are described (Negative Entailment Tests are
> described by reference to this section)
>
> [[
> If the following element is present
> <test:entailmentRules
> rdf:resource="http://www.w3.org/2000/10/rdf-tests/rdfcore/testSche
> ma#simleEntailment" />
> then the test succeeds if the entailment holds according to the rules of
> simple entailment as specified in [RDF-SEMANTICS].
> ]]
>
> According to DanC in discussion post-telecon today, this would satisfy
> his requirements.
>
>
> Consequently, having laid out the options, I'd strongly suggest that the
> WG (well, the chairs) approve the second change as an editorial fix; it
> represents least impact on existing test harnesses, DanC says it
> suffices and it doesn't hurt the WebOnt test cases.
>
>
> With approval of this I can fix the editor's WD (in addition to
> withdrawl of the test cases listed in the actions against me) and have
> the document ready on Monday.
>
> Cheers,
> jan
>
>
> --
> jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/
> Tel +44(0)117 9287088 Fax +44 (0)117 9287112 http://ioctl.org/jan/
> YKYBPTMRogueW... you try to move diagonally in vi.
>
>
Received on Monday, 10 November 2003 05:03:38 EST

This archive was generated by hypermail pre-2.1.9 : Monday, 10 November 2003 05:04:01 EST