Re: SKOS-XL (was RE: SKOS/ synonym provenance (ISSUE-27 AnnotationOnLabel))

Hi Alistair,

Apart from the formal concerns I expressed in my previous mail, I just 
wanted to say that I had also some technical doubts. Mainly regarding 
the correspondence between the "label-as-resource" pattern and the 
"minimal label relation" one: your rules do not consider the attachment 
of the ex:fooRelation to the considered instances of skos:Concept.

This raises again the issue I mentioned once about the minimal label 
relation [4] also lacking a story.
What is the story for contextualizing the "reified" relationship between 
labels? In [4] the relationship resource is linked - via a 
seeLabelRelation property - to the concept to which the labels 
themselves are attached.
I already mentioned the problem in a telecon. If I remember correctly, 
you said that you would attach the reified relationship to each of the 
concepts to which the original literals are attached. This can be 
doable, but I think it might raise some problems one day, and in any 
case miss sound justification. The fact that you forgot it in [2] could 
be a hint :-p

Is it because the problem is not important, contrary to what I think, or 
is there really something?
[And of course this should not hide the fact that the 
"label-as-resource" or "simple extension" lacks a story. Here I agree 
with you...]

Cheers,

Antoine

[4] 
http://www.w3.org/2006/07/SWD/wiki/SkosDesign/RelationshipsBetweenLabels/ProposalFour
[5] http://lists.w3.org/Archives/Public/public-swd-wg/2007Nov/0063.html

> Hi Jon,
>
> You just reminded me, after the amsterdam f2f I wrote up a specification for an *extension module* for SKOS, which I think captures your requirements:
>
> [2] <http://www.w3.org/2006/07/SWD/wiki/SkosDesign/SKOS-XL>
>
> This takes the many-to-one position [3].
>
> My current feeling is *not* to include anything like this in the main SKOS recommendation -- i.e. to limit the SKOS recommendation to *only* dealing with labels as RDF plain literals, which would keep it smaller and simpler. 
>
> I think it would then be quite reasonable to publish something like SKOS-XL as a separate, stand-alone, extension to SKOS, for advanced users. 
>
> The SWDWG could itself publish such an extension, or anyone from the SKOS community could do so. E.g. the FAO used their own extension to represent something like this.
>
> If the SWDWG left it to the community, to help promote discovery and convergence, the SWDWG could set up a wiki page where members of the community could "register" their SKOS extensions ... or we could even use your metadata registry to do that :)
>
> Finally, note that [1] doesn't have any "story" to it -- it's just bare bones. Even as an extension module, [1] would need a story to go with it. To be even considered for inclusion in SKOS proper, it would need a very good story. I haven't got a story at all the moment, and I haven't heard anyone tell one yet either, so my position as stated in the summary of [3] still holds. Have you got a good story? 
>
> Cheers,
>
> Al.
>
> [3] <http://isegserv.itd.rl.ac.uk/public/skos/2007/10/f2f/label-relations.html>
>
> --
> Alistair Miles
> Research Associate
> Science and Technology Facilities Council
> Rutherford Appleton Laboratory
> Harwell Science and Innovation Campus
> Didcot
> Oxfordshire OX11 0QX
> United Kingdom
> Web: http://purl.org/net/aliman
> Email: a.j.miles@rl.ac.uk
> Tel: +44 (0)1235 445440  
>
>   
>> -----Original Message-----
>> From: Jon Phipps [mailto:jonphipps@gmail.com] On Behalf Of Jon Phipps
>> Sent: 20 November 2007 13:17
>> To: Miles, AJ (Alistair)
>> Cc: Antoine Isaac; Daniel Rubin; public-swd-wg@w3.org; Alan Ruttenberg
>> Subject: Re: SKOS/ synonym provenance (ISSUE-27 AnnotationOnLabel)
>>
>> Al,
>>
>> I'd like to suggest in the light of further discussion that 
>> we reconsider Guus's Simple Extension Proposal[1]. Perhaps if 
>> we were able to declare skos:prefLabel as having an 
>> owl:equivalentProperty relationship to the rdfs:label 
>> property of a skos prefTerm, then this would allow us to 
>> effectively join a 'term' graph to a concept by asserting a 
>> typed relationship without impacting the current semantics of 
>> prefLabel. I think this might be far more effective than 
>> simply allowing a resource to be the object of a skos:label property.
>>
>> I believe that Antoine had drawn this pattern on a notepad at 
>> the f2f but it didn't provoke much discussion. As I recall 
>> the main objections to Guus's proposal had to do with 
>> problems with the overloading of 'term' and the fact that 
>> it's subject to rather broad interpretation. Perhaps rather 
>> than simply rejecting the proposal, we could see if we can't 
>> adjust the naming to be more acceptable wrt to the apparent 
>> ambiguity of the term 'term' -- prefLexicalTerm perhaps.
>>
>> Personally I'm far more comfortable allowing the joining of a 
>> term to a concept to both maintain and allow relationships 
>> between terms that can't be effectively expressed with the 
>> more generalizable conceptual relationships supported by skos 
>> than I am with the currently supported solution. It seems to 
>> me that there are far too many instances where publishing a 
>> concept using skos involves enough of a loss of useful data 
>> that it would present a barrier to acceptance of skos.
>>
>> --Jon
>>
>> [1] 
>> http://lists.w3.org/Archives/Public/public-swd-wg/2007May/0057.html
>>
>> On Nov 20, 2007, at 7:40 AM, Miles, AJ ((Alistair)) wrote:
>>
>>     
>
>
>

Received on Friday, 23 November 2007 22:33:59 UTC