RE: Datatype (Map) Conformance Strangeness

Hi!

I still believe that we should avoid talking about "datatype maps" in the
Conformance document, but should talk about "sets of datatypes", instead. I
think that it doesn't provide any relevant information to readers of this
document to talk about datatype maps, and it distracts from the really
relevant bits. Talking about a datatype map instead of a set of datatypes is
a bit like talking about the bottle, if one really wants to talk about the
wine inside. A datatype map is more or less just a container for a set of
datatypes, and the included datatypes are what really counts when we talk
about conformance.

In particular, I consider it to be more confusing than helpful to say in
Conformance that there are two notions of a datatype map in our spec. This
may sound to people as if the set of datatypes supported by the RDF-Based
Semantics is different from the set of datatypes supported by "OWL 2".
Neither is this the case nor should it be the case (modulo rdf:XMLLiteral,
perhaps).

What is actually different are the two definitions of a datatype /map/. But
my understanding is that a datatype map is mainly an "adapter" to "connect"
a /set/ of datatypes (for example the OWL 2 datatypes) to an interpretation
of either the RDF-Based Semantics or the Direct Semantics. And since the
definitions of an interpretation are different for the two semantics, it
shouldn't be too surprising that the definitions of a datatype /map/ are (at
least slightly) different, either.

Nevertheless, there are a still few points, which probably should be
changed:

* What has been called "OWL 2 Full Datatype Map" is now called "OWL 2
RDF-Based Datatype Map"
  <http://www.w3.org/2007/OWL/wiki/RDF-Based_Semantics#def-owldatatypemap>

* Maybe we should at least rename the title of the section to "Datatype
Conformance" instead of "Datatype Map Conformance".

* It is not really true that the "OWL 2 Datatype Map" is defined in Section
4 of the Structural Spec. This section primarily only specifies the /set/ of
datatypes of OWL 2 and its properties. There is only a slight hint
concerning datatype /maps/ in the beginning of this section (and I wouldn't
be unhappy to see it completely removed). The actual definition of the
datatype /map/ is in Section 2.1 of the Direct Semantics:
<http://www.w3.org/2007/OWL/wiki/Semantics#Vocabulary>.

Michael

>-----Original Message-----
>From: Ian Horrocks [mailto:ian.horrocks@comlab.ox.ac.uk]
>Sent: Monday, April 13, 2009 4:50 PM
>To: Michael Schneider
>Cc: public-owl-wg@w3.org
>Subject: Re: Datatype (Map) Conformance Strangeness
>
>I agree with Peter that this is a technical document and not one that
>I would expect any "user" to be interested in -- it is mainly
>targeted at implementers of OWL tools who should, I assume, be
>familiar with terms such as datatype map. So, I don't see mentioning
>datatype maps here to be inherently problematical.
>
>Currently, conformance is the *only* place that specifies what the
>datatype map has to be when the direct semantics is applied. We can
>argue about the wisdom of this, but I suggest not to try to change it
>before LC as there is too much danger that we will just mess things
>up -- we could always re-organise things later as the change would
>only be editorial. This being the case, we can't eliminate mention of
>datatype maps from Conformance. I left this statement mentioning both
>the OWL Datatype map and the OWL 2 Full Datatype Map as it would look
>strangely asymmetrical otherwise, and I don't believe that this
>statement is actually wrong (even if it is redundant).
>
>However, it is certainly the case that the note about the datatypes
>that can occur in conformant OWL DL documents is in the wrong place
>and should *not* refer to datatype maps but rather to the set of
>datatypes listed in Section 4 of SS&FS (plus rdfs:Literal). I have
>therefore moved this note into the Document Conformance section.
>
>It is also the case that having Datatype Conformance be a subsection
>of Document conformance is wrong, as the former talks about semantic
>conditions and the latter about syntactic ones. I therefore promoted
>Datatype conformance into its own subsection.
>
>I think that the  above mentioned changes are sufficient for the time
>being -- as I said above, we can think some more about the
>organisation of these various documents after LC.
>
>Ian
>
>
>On 12 Apr 2009, at 08:46, Michael Schneider wrote:
>
>> Hi Ian!
>>
>> I argued for not talking about "datatype maps" in the Conformance
>> document
>> at all. I suggest to just talk about "(sets of) datatypes". My
>> proposed
>> revision of the section in my previous mail reflects this.
>>
>> I don't see why referring to datatype maps would be necessary or would
>> provide any relevant additional information. We will just open the
>> Conformance document up to unnecessary criticism.
>>
>> I consider datatype maps as an internal aspect of the Direct
>> Semantics and
>> the RDF-Based Semantics. So let's talk about datatype maps
>> exclusively in
>> the semantics documents. I think it would even be best to not talk
>> about
>> datatype maps in the Structural Spec, but, again, only about (sets of)
>> datatypes.
>>
>> Michael
>>
>>> -----Original Message-----
>>> From: Ian Horrocks [mailto:ian.horrocks@comlab.ox.ac.uk]
>>> Sent: Saturday, April 11, 2009 12:42 AM
>>> To: Michael Schneider
>>> Cc: public-owl-wg@w3.org
>>> Subject: Re: Datatype (Map) Conformance Strangeness
>>>
>>> I agree with you that this has got rather confused. I think that the
>>> problem is twofold:
>>>
>>> 1) I added the (redundant) note about conformant ontology documents
>>> in the wrong place -- this could actually be part of the definition
>>> of an OWL 2 DL ontology document (it is redundant because the
>>> condition is already one of the conditions that an ontology must
>>> satisfy in order to be an OWL 2 ontology as specified in Section 3 of
>>> SS&FS).
>>>
>>> 2) Section 2.1.2 is talking about semantic conditions, yet it is in
>>> the "Document Conformance section.
>>>
>>> Thus, I think that the correct way to fix the problem is:
>>>
>>> 1) Move the note on datatypes to be part of the definition of an OWL
>>> 2 DL ontology document (or get rid of it altogether).
>>>
>>> 2) Promote 2.1.2 to (sub) section 2.2 (Tool Conformance will then
>>> become Section 2.3).
>>>
>>> I also think that the text should be changed slightly to say:
>>>
>>> "In OWL 2, semantic conditions are defined with respect to a datatype
>>> map. This MUST be either the OWL 2 datatype map (as defined in
>>> Section 4 of the OWL 2 Syntax specification [OWL 2 Specification]),
>>> an OWL 2 RDF-Based datatype map (as defined in Section 4.1 of the OWL
>>> 2 RDF-Based Semantics [OWL 2 RDF-Based Semantics]), or an extension
>>> of the OWL 2 datatype map to include additional datatypes.
>>>
>>> OWL 2 Profiles may support only a reduced set of datatypes. This is,
>>> however, a syntactic condition that must be met by documents in order
>>> to fall within the relevant profile, and the semantic conditions on
>>> the supported datatypes are unchanged, i.e., they are still defined
>>> by an OWL 2 RDF-Based datatype map or a (possibly extended) OWL 2
>>> datatype map. These datatype maps define semantic conditions on
>>> unsupported datatypes, but as these datatypes never occur in
>>> conforming documents the additional conditions are simply
>>> irrelevant."
>>>
>>> I assume that it is correct to say that semantic conditions may be
>>> defined by an OWL 2 RDF-Based datatype map -- presumably tools using
>>> the RDF-Based semantics will use such a datatype map.
>>>
>>> Ian
>>>
>>>
>>>
>>> On 10 Apr 2009, at 16:19, Michael Schneider wrote:
>>>
>>>> Hi!
>>>>
>>>> I had a closer look at the "Datatype map conformance" section
>>>> (§2.1.2) in the Conformance document:
>>>>
>>>>   <http://www.w3.org/2007/OWL/wiki/index.php?
>>>> title=Conformance&oldid=21801#Datatype_Map_Conformance>
>>>>
>>>> I am pretty confused by the current state. I don't understand why
>>>> the section refers to the OWL 2 Full datatype map, or to datatype
>>>> maps at all? The section is still about syntactic conformance, and
>>>> the only relevant thing here seems to be which datatypes may occur
>>>> in ontologies.
>>>>
>>>> I think, the paragraph confuses two things:
>>>>
>>>> 1) The set of datatypes and their properties, i.e. value spaces,
>>>> lexical spaces, facets. These are specified in the Structural Spec
>>>> (mainly by referring to XSD and other specifications) and are
>>>> invariant for the Direct Semantics and the RDF-Based Semantics.
>>>>
>>>> 2) The definitions of datatype maps. These definitions are part of
>>>> the two semantics, and they differ from each other structurally in
>>>> order to match the different semantic frameworks.
>>>>
>>>> I believe only 1) is relevant for Section 2.1.2, while the
>>>> (different) aspects of datatype maps in 2) have no relevance for
>>>> syntactic conformance at all.
>>>>
>>>> Maybe the confusion already stems from the title that has been
>>>> chosen for this section (and has been around for a while, I think):
>>>> I'd say that it should be changed from "Datatype Map Conformance"
>>>> to "Datatype Conformance", because datatype /maps/ do not really
>>>> play a role here, only the /set/ of datatypes supported by OWL 2 is
>>>> of relevance.
>>>>
>>>> Here is a proposal for a revision of the Section as I think it
>>>> would be more appropriate:
>>>>
>>>>>>>>>>>>> BEGIN PROPOSAL <<<<<<<<<<<<<<<
>>>>
>>>> ==== Datatype Conformance ====
>>>>
>>>> In OWL 2, semantic conditions are defined with respect to a set of
>>>> datatypes. This <em title="MUST in RFC 2119 context"
>>>> class="RFC2119">MUST</em> be either the set of datatypes as defined
>>>> in [[Syntax#Datatype_Maps|Section 4]] of the OWL 2 Syntax
>>>> specification [<cite>[[#ref-owl-2-specification|OWL 2
>>>> Specification]]</cite>]), or an extension of this set to include
>>>> additional datatypes.
>>>>
>>>> Note that:
>>>> # A conformant OWL 2 DL ontology document <em title="MUST NOT in
>>>> RFC 2119 context" class="RFC2119">MUST NOT</em> use datatypes other
>>>> than those specified in [[Syntax#Datatype_Maps|Section 4]] of the
>>>> OWL 2 Syntax specification [<cite>[[#ref-owl-2-specification|OWL 2
>>>> Specification]]</cite>].
>>>> # OWL 2 Profiles may support only a reduced set of datatypes. This
>>>> is, however, a syntactic condition that must be met by documents in
>>>> order to fall within the relevant profile, and the semantic
>>>> conditions on the supported datatypes are unchanged. This also
>>>> defines conditions on unsupported datatypes, but as these datatypes
>>>> never occur in conforming documents the additional conditions are
>>>> simply irrelevant.
>>>>
>>>>>>>>>>>>> END OF PROPOSAL <<<<<<<<<<<<<<<<<
>>>>
>>>> Cheers,
>>>> Michael
>>
>> --
>> Dipl.-Inform. Michael Schneider
>> Research Scientist, Dept. 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
>> ======================================================================
>> =
>>

--
Dipl.-Inform. Michael Schneider
Research Scientist, Dept. 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
=======================================================================