W3C home > Mailing lists > Public > public-swd-wg@w3.org > December 2008

Re: ISSUE-157 Draft response [was Re: OWL WG LC comment for SKOS reference document]

From: Alistair Miles <alistair.miles@zoo.ox.ac.uk>
Date: Tue, 9 Dec 2008 14:34:58 +0000
To: Antoine Isaac <aisaac@few.vu.nl>
Cc: Guus Schreiber <schreiber@cs.vu.nl>, public-swd-wg@w3.org
Message-ID: <20081209143455.GA22792@skiathos>

Hi Antoine,

First let me say I like the story the Primer tells about documentation
properties, would like to keep it that way if at all possible.

So given the primer's position, we have to try to find a way to
satisfy Peter w.r.t. the data model.

Personally, I could live with it if we keep the Primer as is and
implement the changes proposed by Guus. It would not be ideal, but I
could live with it, especially as I expect most SKOS developers will
come to SKOS through the Primer, and will not be concerned with
description logics compatibility.

I'm now trying to think of alternatives. 

It might be an option to make all the documentation properties
instances of owl:AnnotationProperty, rather than
owl:ObjectProperty. But then we have the sub-property axioms to worry
about. From [1] I see this may be acceptable for OWL 2: "There are
only three axioms that can be used on annotation properties:
AnnotationPropertyDomain, AnnotationPropertyRange and
SubAnnotationPropertyOf axioms. These special annotation axioms have
no semantics in OWL DL, but the normal semantics in OWL Full via their
mapping to the standard RDF vocabulary." I guess this would be a
moderately substantial change. I could live with this.

The only other alternative I can think of is to go back to a single
usage pattern, with a dedicated property to carry the note content
instead of rdf:value, e.g. ex:MyConcept skos:note [ skos:noteContent
"foo bar"@en; ]. But that would be a rather substantial change to both
Reference and Primer, and not backwards-compatible with the majority
of extant data. I don't think I could support this.

I suspect that, even if we change the SKOS documentation properties to
instance of owl:AnnotationProperty, Peter will still object to
rdf:value. So there may not be any way to get the SKOS Reference and
the SKOS Primer to mirror each other perfectly. We may have to find
some sort of compromise.

Cheers,

Alistair

[1] http://www.w3.org/TR/2008/WD-owl2-new-features-20081202/#Extended_annotations

On Thu, Dec 04, 2008 at 11:24:55PM +0100, Antoine Isaac wrote:
> Hi Guus, Alistair,
>
> I'm ok for many of the aspects of this answer.
>
> I am however **especially worried** with the following suggestions:
>
>> - move examples 24 and 25 to Sec. 7.5.2., i.e. the note on alignment of the range semantics with OWL semantics; this would leave only the "OWL 2 DL"-consistent example in Sec. 7.4;
>>
>> - change example 27 to use as subject a resource with an rdf:value property;
>
> The current section 7.4 has the positive point for the moment that it examplifies notes in what I feel is an order of increasing complexity and decreasing usefulness. I indeed expect that an overwhelming proportion of KOSs require just plain literal notes. And even if some may require structured resources, I think then that many users would still prefer to avoid coining URIs for all of them.
>
> My worry is that shuffling the examples around would give SKOS users a wrong message: the first example presented would be the most complex and rare one. And the more useful ones would appear in a kind of footnote, with warnings around that would concretely present them as the bad boys that make the SKOS ontology OWL Full.
> This I'm afraid strongly encourages to adapt something that is not as simple, even when it's not needed by the application at hand.
>
> I understand Peter's worries, but (my?) story is that skos:note is rather a datatypeProperty which was made into an objectProperty to fit more sophisticated (but, of course, important) cases, rather than the opposite. I would find it a pity that this story remains only in the Primer -- see the respective positions of sections 2.4 and 4.2 [1] -- which has a much lower normative power.
>
> Antoine
>
> [1] http://www.w3.org/TR/skos-primer/
>
>
>> Hi Guus,
>>
>> This looks good to me.
>>
>> Thanks,
>>
>> Alistair.
>>
>> On Thu, Dec 04, 2008 at 03:23:28PM +0100, Guus Schreiber wrote:
>>> All,
>>>
>>> This is a *draft response* wrt ISSUE-157. Please send comments to the 
>>> list. As agreed during the last telecon, we intend to send the actual 
>>> response Friday 5 Dec, end of the day, so reactions before that time  
>>> would be very much appreciated.
>>>
>>> Guus
>>>
>>>
>>> -----------------------
>>> Peter,
>>>
>>> Thanks again for your comments on behalf of the OWL WG [1]. This is a 
>>> response to part of these comments, marked in our issues list as  
>>> ISSUE-157 "SKOS and OWL 2 analysis" [2]:
>>>
>>> [[
>>>   The OWL WG notes that some parts of the SKOS specification and some 
>>> examples in the reference document do not fit within OWL 2 DL and 
>>> that thus may not be fully supported by Semantic Web tools.  The OWL 
>>> WG presents the following analysis of the SKOS specification and 
>>> examples, to indicate where representation capabilities beyond OWL 1 
>>> DL are used. The OWL WG further notes that in many cases the SKOS 
>>> specification fits  within OWL 2 DL, but that the examples do not.  
>>> The OWL WG suggests removing those examples that do not fit within 
>>> OWL 2 DL.([from [1])
>>> ]]
>>>
>>> below you find our responses to the SKOS aspects that are not OWL 2 
>>> DL compliant. As a general strategy, we have tried as much as 
>>> possible to accommodate the alignment with OWL 2 DL. A number of 
>>> specific points cannot be resolved at this time (see below), so we 
>>> have decided to POSTPONE this issue.
>>>
>>> [[
>>> Section: Lexical Labels
>>> Language: OWL 2 Full
>>> Issue: subproperty of rdfs:label
>>> Suggestion: don't use rdfs:label
>>> ]]
>>>
>>> We prefer to keep the subProperty relation; however, we propose to 
>>> change the type of the lexical label to owl:AnnotationProperty. 
>>> Assuming that OWL 2 DL will support subproperty statements between 
>>> annotation properties, this change shroud at least partially solve 
>>> the issue.
>>>
>>> [[
>>> Section: Lexical Labels
>>> Language: OWL 2 Full
>>> Issue: objects as values of data property (example)
>>> Suggestion: don't do this
>>> ]]
>>>
>>> We assume you refer to example 17; we propose to remove this example.
>>>
>>> [[
>>> Section: Documentation
>>> Language: OWL 2 Full
>>> Issue: using literal in object property (examples)
>>> Suggestion: don't do this
>>>
>>> Section: Documentation
>>> Language: OWL 2 Full
>>> Issue: use of rdf:value (example)
>>> Suggestion: don't use rdf:value
>>> ]]
>>>
>>> We assume you refer to examples 24,25 and 27. We propose the 
>>> following changes:
>>>
>>> - move examples 24 and 25 to Sec. 7.5.2., i.e. the note on alignment 
>>> of the range semantics with OWL semantics; this would leave only the 
>>> "OWL 2 DL"-consistent example in Sec. 7.4;
>>>
>>> - change example 27 to use as subject a resource with an rdf:value  
>>> property;
>>>
>>> - reverse the order of the two notes, i.e. Sec. 7.5.1. and Sec. 7.5.2.
>>>
>>> We prefer to keep the rdf:value examples, as this the RDF standard  
>>> defines this as the preferred way of identifying the value in "value" 
>>> resources. However, these rdf;Value are now only used in "Note" 
>>> sections, which contain the appropriate caveats about the semantics.
>>>
>>> [[
>>> Section: Lexical Labels
>>> Language: not OWL
>>> Issue: axiom schema for unique prefLabel
>>> Suggestion: include qualified cardinality restrictions only
>>>   for languages used (defined using datatype restrictions)
>>>
>>> Section: Concept Collections
>>> Language: OWL 2 Full
>>> Issue: ordering with typing
>>> Suggestion: see [1]
>>>
>>> Section: SKOS XL
>>> Language: OWL 2 Full
>>> Issue: data property chains
>>> Suggestion: ??
>>> ]]
>>>
>>> We assume these three issues refer to constraints S14 (lexical 
>>> labels), S35 (ordered collections) and S56, S57 & S58 (SKOS XL). 
>>> Indeed, these constraints can (currently) not be expressed in OWL. 
>>> However, these are useful constraints for tool developers and we 
>>> therefore prefer to keep these in the SKOS Reference.
>>>
>>>
>>> Please let us know whether you can live with this response.
>>>
>>> Thanks again for your comments!
>>>
>>> Regards,
>>> Guus
>>>
>>> [1] http://lists.w3.org/Archives/Public/public-swd-wg/2008Oct/0059.html
>>> [2] http://www.w3.org/2006/07/SWD/track/issues/157
>>
>>> begin:vcard
>>> fn:Guus Schreiber
>>> n:Schreiber;Guus
>>> org:VU University Amsterdam, Computer Science
>>> email;internet:schreiber@cs.vu.nl
>>> title:Prof. dr. x-mozilla-html:FALSE
>>> url:http://www.cs.vu.nl/~guus/
>>> version:2.1
>>> end:vcard
>>>
>>
>>
>

-- 
Alistair Miles
Senior Computing Officer
Image Bioinformatics Research Group
Department of Zoology
The Tinbergen Building
University of Oxford
South Parks Road
Oxford
OX1 3PS
United Kingdom
Web: http://purl.org/net/aliman
Email: alistair.miles@zoo.ox.ac.uk
Tel: +44 (0)1865 281993
Received on Tuesday, 9 December 2008 14:35:46 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:07:55 UTC