RE: OWL2 RL - conformance test question(s)

[Maybe Markus or Mike can help? We are talking about the flag in the OWL 2
Test Wiki that was introduced to indicate tests for the OWL 2 RL/RDF rules.]

>-----Original Message-----
>From: Barry Bishop [mailto:barry.bishop@ontotext.com]
>Sent: Saturday, May 29, 2010 12:50 PM
>To: Michael Schneider
>Cc: public-owl-dev@w3.org
>Subject: Re: OWL2 RL - conformance test question(s)
>
>Hi Michael,
>
>Thanks very much for the quick response.
>
>To be honest, I'm not really sure what to do next. My rule-engine and
>rule-set seem to have the correct behaviour, because I have tested
>against a set of unit tests of my own creation. What I was really hoping
>to do, was to get hold of official/approved tests that could thoroughly
>test my implementation.
>
>Do you suppose that 'rlrulescomplete=Yes' is used consistently
>throughout the test descriptions? 

I was thinking that I can answer this question by a simple "Yes" and give
you a normative pointer. I remember that the WG needed some way to indicate
that a test works under the OWL 2 RL/RDF Rules, without the requirement to
also fall into the OWL 2 RL syntactic fragment (the OWL 2 RL/RDF rules are
defined for arbitrary [even generalized] RDF graphs). At some time, this
flag was introduced, and you can see its effect in the Test Wiki by the
sentence:

    This test is not in OWL 2 RL but can be passed 
    by implementations that use the OWL RL derivation rules.

(By "not in OWL 2 RL" the syntactic fragment is meant.) But I cannot find
where this decision has been written down (except for the obviously existing
translation code in the Wiki). 

I would have expected to find some discussion about it with a simple search
in the OWL WG mailing list, but failed. 

I also do not find a corresponding definition in the Test Ontology, which is
described in the "Conformance" document. In fact, I do not even see the flag
being translated into some corresponding ontology element, when looking at
the translation of the test case into OWL (available by the "[Download OWL]"
link in the upper right of each test case). It might be represented by some
combination of different elements, otherwise I would consider this a missing
feature in the ontology.

I guess that the WG has decided to introduce this flag in one of their
weekly teleconferences, but checking this out would take me much time.

So I hope that Markus or Mike can help out, because they were mainly working
on the test Wiki and the Ontology.

Best,
Michael

>If so, I could extract them one by one.
>
>Thanks a lot,
>barry
>
>On 29/05/2010 00:11, Michael Schneider wrote:
>> [asked to the OWL WG, but posted to OWL-DEV, so no need for a formal
>reply]
>>
>> Hi Barry!
>>
>> These are clearly not tests for the OWL 2 RL/RDF Rule language, they
>are
>> much too hard.
>>
>> Here is an example for a test that is really intended for the RL
>Rules:
>>
>>
>> http://owl.semanticweb.org/page/Rdfbased-sem-restrict-somevalues-inst-
>subj
>>
>> When you view the source code of this test, you will find the
>following line
>> at the end:
>>
>>     rlrulescomplete=Yes
>>
>> This is the indicator for those tests that are created for the RL
>Rules.
>>
>> There has been a specific bulk download of only these tests in the
>past, but
>> it looks to me that it is broken now. On the right hand side of the
>test
>> homepage, you can find the link
>>
>>     Tests for RL-RDF-rule reasoners
>>
>> but it is empty.
>>
>> You may play around with SMW queries yourself. But a quick shot would
>be to
>> simply hit the link to my name under "Top Contributors", which will
>give you
>> a list of tests that a colleague and I have contributed, and these
>tests
>> have all been created specifically for the RL rules.
>>
>> You can also get a larger test suite, including all these tests, but
>also
>> tests concerning the different datatypes of OWL 2 RL (and a few more
>> aspects). Get our OWLED paper at [1]. It contains a download link in
>the
>> references. If you have any further questions about this test suite,
>feel
>> free to ask me privately.
>>
>> Best,
>> Michael
>>
>> [1] Michael Schneider and Kai Mainzer:
>>     A Conformance Test Suite for the OWL 2 RL/RDF Rules Language
>>     and the OWL 2 RDF-Based Semantics.
>>
><http://www.webont.org/owled/2009/papers/owled2009_submission_19.pdf>
>>
>>> -----Original Message-----
>>> From: public-owl-dev-request@w3.org [mailto:public-owl-dev-
>>> request@w3.org] On Behalf Of Barry Bishop
>>> Sent: Friday, May 28, 2010 6:52 PM
>>> To: public-owl-dev@w3.org
>>> Subject: OWL2 RL - conformance test question(s)
>>>
>>> Hello OWL2 working group,
>>>
>>> I would like to ask a couple of fairly straightforward questions
>>> regarding the conformance tests, and the semantics, of the RL
>profile.
>>>
>>> It is very likely that I have missed something fundamental, so I
>would
>>> be very pleased if someone could nudge me in the right direction.
>>>
>>> I have implemented the RDF semantics using the rule set given in:
>>> http://www.w3.org/TR/owl2-
>>> profiles/#Reasoning_in_OWL_2_RL_and_RDF_Graphs_using_Rules
>>>
>>> However, I am finding this hard to reconcile with the conformance
>tests
>>> found here:
>>> http://owl.semanticweb.org/exports/approved/profile-RL.rdf
>>>
>>> because some of the expected conclusions of the positive entailment
>>> tests can not by produced by any of the RL entailment rules (I have
>been
>>> careful to select only those tests labelled with 'test:semantics
>>> test:RDF-BASED').
>>>
>>> A couple of specific examples:
>>>
>>>
>>> Chain2trans
>>> ===========
>>> Premise ontology:
>>>
>>> <http://example.org/#p> rdf:type owl:ObjectProperty .
>>> _:bnode0 rdf:first <http://example.org/#p> ;
>>>    rdf:rest _:bnode1 .
>>> _:bnode1 rdf:first <http://example.org/#p> ;
>>>    rdf:rest rdf:nil .
>>> <http://example.org/#p> owl:propertyChainAxiom _:bnode0 .
>>>
>>> Conclusion ontology:
>>>
>>> <http://example.org/#p> rdf:type owl:TransitiveProperty .
>>>
>>> Search as I might, I can not find any entailment rules with
>>> owl:TransitiveProperty in the head. So how can this be?
>>>
>>>
>>> DisjointClasses-001
>>> ===================
>>> Premise ontology:
>>>
>>> <http://example.org/Boy> rdf:type owl:Class .
>>> <http://example.org/Girl> rdf:type owl:Class .
>>> <http://example.org/Boy> owl:disjointWith <http://example.org/Girl> .
>>> <http://example.org/Stewie> rdf:type <http://example.org/Boy> .
>>>
>>> Conclusion ontology:
>>>
>>> <http://example.org/Girl> rdf:type owl:Class .
>>> _:bnode1 rdf:type owl:Class ;
>>>    owl:complementOf <http://example.org/Girl> .
>>> <http://example.org/Stewie> rdf:type _:bnode1 .
>>>
>>> However, there are no OWL2-RL rules that have blank nodes in the
>head.
>>>
>>> Perhaps I have misunderstood this (in regard to RDFS semantics):
>>>
>>> "An OWL 2 RL/RDF implementation MAY include these triples and
>entailment
>>> rules as necessary without invalidating the conformance requirements
>for
>>> OWL 2 RL [OWL 2 Conformance]."
>>>
>>> - my implementation does not have, for example, RDF(S) se1 & se2
>rules.
>>>
>>> Or is there some other required behaviour somewhere that I have
>>> overlooked?
>>>
>>> Many thanks in advance,
>>> barry

--
Dipl.-Inform. Michael Schneider
Research Scientist, Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: michael.schneider@fzi.de
WWW  : http://www.fzi.de/michael.schneider
=======================================================================
FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts, Az 14-0563.1, RP Karlsruhe
Vorstand: Prof. Dr.-Ing. Rüdiger Dillmann, Dipl. Wi.-Ing. Michael Flor,
Prof. Dr. Dr. h.c. Wolffried Stucky, Prof. Dr. Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
=======================================================================

Received on Saturday, 29 May 2010 14:02:51 UTC