Re: ISSUE-130 / ACTION-194 Come up with a proposal for conformance

Thanks for the thoughtful review. Some responses in line below...

On 2 Sep 2008, at 09:48, Michael Schneider wrote:

> [related to ISSUE-130 and ISSUE-131 (Conformance and OWL RL  
> Unification)]
>
> Hi!
>
> I have read the Conformance proposal at
>
>   <http://www.w3.org/2007/OWL/wiki/Conformance>
>
> It's a clearly written document. I have several comments/questions.
>
> * 1.1.1.
>
>   """
>   Any RDF/XML document is an OWL 2 Full ontology document.
>   """
>
> I suggest to replace "RDF/XML document" by "RDF document" or "RDF  
> graph".
> Whether something is a Full ontology should not depend on a  
> specific RDF
> serialization.

I agree. This was a hangover from OWL 1. I used RDF document, because  
conformance relates to concrete artefacts such as documents not  
abstract objects such as RDF graphs.

>
> * 1.1.1.
>
>    """
>    An OWL 2 Full ontology document is an OWL 2 RL
>    ontology document iff it can be successfully
>    parsed using the canonical RDF parsing process
>    and the resulting ontology in the functional-style
>    syntax satisfies all the restrictions on OWL 2 RL ontologies
>    """
>
> Now I understand that one of the purposes of the (so far) OWL RL DL  
> syntax
> is to define /syntactic/ conformance. For example, the following  
> RDF graph
> (aka OWL 2 Full ontology) is /not/ a syntactically conformant OWL RL
> ontology:
>
>   ex:C rdfs:subClassOf _:x
>   _:x owl:unionOf (SEQ ex:D1 ex:D2)
>
> because, after the reverse mapping, it would not meet the RL syntax
> restrictions. I'm personally fine with this.
>
> * 1.2.1.
>
>   """
>   An OWL 2 entailment checker takes as input
>   two OWL 2 ontology documents O1 and O2
>   """
>
> Isn't it better to talk about "two /imports-closed collections of/  
> OWL 2
> ontology documents"?

I did think about making the input to entailment checkers be  
something that would abstract away from imports related issues and  
perhaps even from parsing issues. I decided not to do this, because  
it would solve one problem by introducing another. Computing the  
imports closure is quite a complex process, and is dependent on  
several factors, including document locations. If we were going to do  
as you suggest here, then I think we would need to define something  
like an "OWL 2 imports-closure checker", and define comformance  
criteria for that. In the end, it seems simpler to bundle all of this  
into the definition of a single artefact.


>
> * 1.2.1.
>
>   """
>   and either the Model-Theoretic Semantics [OWL 2 Semantics]
>   or the RDF-Based Semantics [OWL 2 RDF-Based Semantics]
>   """
>
> I just want to state that the RDF-Based Semantics is also a model- 
> theoretic
> semantics, so this is not a useful distinction criterion.

These are what the documents are currently called, so in that sense  
it is a very useful distinction. You may want to have a discussion  
about these names.

>
> * 1.2.1
>
>   """
>   An OWL 2 entailment checker returns one word,
>   being True, False, or Unknown.
>   """
>
> Perhaps, it's better to talk about "result" instead of "word".

I agree. I changed the document accordingly.

>
> * 1.2.1
>
>   """
>   An OWL 2 Full entailment checker is an OWL 2 entailment checker
>   that takes RDF/XML documents as input
>   """
>
> Again "RDF/XML". But here, one can simply write "OWL 2 Full ontology
> documents", because this term has already been defined elsewhere,  
> and it
> would also be better in line with the rest of the "entailment checker"
> definitions in this section.

I refer to RDF here as it is more direct -- the definition of an OWL  
2 Full ontology document is just that it is an RDF document. I wonder  
if we need to define OWL 2 Full ontology document at all?

>
> * 1.2.1 Analog for "OWL 2 RL entailment checker". But my question  
> is: Should
> it be "OWL 2 /Full/ ontology documents" or "OWL 2 /RL/ ontology  
> documents"?

It should be OWL Full ontology document / RDF document.


>
> * 1.2.1
>
>   """
>   An OWL 2 RL entailment checker
>   [...] MUST return True only when O1 entails O2,
>   and it must return False
>   only when FO(O1) ? R does not entail FO(O2)
>   under the standard first-order semantics
>   """
>
> Just to be clear: This means that there can be two conformant OWL 2 RL
> entailment checkers C1 and C2, where for the same query "G_L |=  
> G_R ?" C1
> answers "False" (because G_R does not follow from G_L via the  
> rules), while
> C2 answers "True" (because G_R follows from G_L via the Full  
> semantics). I
> suggest that we state it explictly that this is intended. Otherwise  
> people
> might get confused and think that this is a bug.

This could be explicitly stated. We could also/instead strengthen the  
statement about warnings when the input ontologies are such that this  
could arise -- currently implementations MAY issue such a warning;  
this could be strengthened to SHOULD or MUST.


>
> * 1.2.1: "terminating and complete entailment checker"
>
>   """
>   reject the input as syntactically invalid
>   """
>
> I don't understand what this means in the case of OWL RL. According to
> 1.1.1, RDF seems to be the reference syntax, but RDF graphs are  
> only said to
> be "RL conformant", if they can be successfully parsed into  
> Functional Style
> syntax and satisfy the RL restrictions. So I wonder whether all RDF  
> graphs
> or only those matching the RL restrictions (the "syntactically OWL RL
> conformant ontology documents"!) are syntactically valid input to  
> an OWL RL
> entailment checker. Perhaps it is meant that an OWL RL entailment  
> checker
> MAY reject syntactically non-conformant OWL 2 RL documents, but if it
> accepts them, then it has to provide the True/False/Unknown  
> behavior as
> stated (e.g. it must not say "True", if the OWL Full semantics do not
> entail). I would like to see some clarification here.

This is a general statement covering all checkers. In the case of OWL  
RL checkers, it is clearly stated that they take RDF documents as  
input, so I don't see any confusion there. Input could be rejected as  
syntactically invalid if, e.g., it isn't and RDF graph.


>
> * 1.2.1:
>
>   """
>   An OWL 2 entailment checker is terminating if,
>   given sufficient memory resources
>   """
>
> I think that two slight clarifications are needed here:
>
>   (1) Is by "given sufficient memory resources" meant that the  
> amount of
> memory is determined "a priorily" by the problem size? Or that  
> memory can be
> appended on demand, such as in the case of a printer running out of  
> paper?

I don't see the problem with this statement -- nothing needs to be  
said (or should be said) about procedural mechanisms for determining  
or allocating memory.

>
>   (2) Definition of "completeness": Does this definition allow an  
> input on
> which the entailment checking process does not terminate? Or is the
> definition of termination already part of the definition of  
> completeness?

No. Many algorithms used in theorem proving are sound and complete  
but non-terminating.

Ian


>
> Best,
> Michael
>
>> -----Original Message-----
>> From: Ian Horrocks
>> Sent: Monday, August 25, 2008 1:27 PM
>> To: public-owl-wg Group WG
>> Subject: ISSUE-130 / ACTION-194 Come up with a proposal for  
>> conformance
>>
>>
>> A draft proposal is now available in the Wiki [1]. The idea is that
>> this would become the conformance section in the Test document [2].
>>
>> Ian
>>
>> [1] http://www.w3.org/2007/OWL/wiki/Conformance
>> [2] http://www.w3.org/2007/OWL/wiki/Test#Conformance
>
> --
> Dipl.-Inform. Michael Schneider
> FZI Forschungszentrum Informatik Karlsruhe
> Abtl. Information Process Engineering (IPE)
> Tel  : +49-721-9654-726
> Fax  : +49-721-9654-727
> Email: Michael.Schneider@fzi.de
> Web  : http://www.fzi.de/ipe/eng/mitarbeiter.php?id=555
>
> 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 Regierungspräsidium Karlsruhe
> Vorstand: Rüdiger Dillmann, Michael Flor, Jivka Ovtcharova, Rudi  
> Studer
> Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
>

Received on Tuesday, 2 September 2008 16:55:45 UTC