Re: [SKOS] About notations and datatypes (was Re: Comments on Vocabularies document, v 1.15)

Norman,
>
> A slightly broader point:  We were hoping to get the Vocabularies 
> document to an IVOA Recommendation within the next month or so, thus 
> before the SKOS document itself was due to become a W3C REC.  We 
> justified this questionable practice to ourselves because we were 
> using only core features of SKOS, unlikely to change between the 
> current drafts and the REC.  However I see from the list of issues 
> which have appeared on the SWD list, that things like skos:notation 
> are still to some extent contestable [1].  This is making me worry 
> that perhaps we should hold back after all.
>
> Do you think that there is likely to be significant change to the sort 
> of core SKOS aspects we've refererred to in our document?

Most of the features you use are quite stable, indeed thay are present 
as such since the 2005 version. And even if there is a change in the 
semantics of these properties/classes, this may only affect your 
document very marginally and quite certainly, not at all -- there is no 
mention of cycle constraints, for instance, just transitivity.
The only issues could be with notations and mapping. But if the 
datatypes in notation may raise issues (which is formally not the case 
now!), your use of mapping properties does not seem to me likely to 
suffer any problems. Plus you give an explicit warning to the reader in 
the related section of your doc...


>   Or, a related question, is end-December 2008 still the expected date 
> for SKOS to become a REC?  (If so, it's close enough that the argument 
> for going ahead out of sync starts to evaporate).

I was not attending the last telecon, but I don't think the scehdule has 
been changed :-)

Best,

Antoine

>
> Best wishes,
>
> Norman
>
>
> [1] http://www.w3.org/2006/07/SWD/track/issues/184
>
>
>
>
>
> On 2008 Oct 4, at 14:26, Antoine Isaac wrote:
>
>>
>> Hello Norman,
>>
>> Glad that my comments on IVOA vocabularies document could help a bit!
>>
>> About the following issue:
>>
>>>
>>>> - I think your proposed example for notations is not compliant with 
>>>> what the SKOS Reference and Primer say about these [1,2]. 
>>>> skos:notation should be used with literals with a specific 
>>>> datatype, as in
>>>> ex:udc512 skos:prefLabel "Algebra" ;
>>>> skos:notation "512"^^ex:myUDCNotationDatatype .
>>>> (and yes, it is not a very simple representation. Which is why we 
>>>> left the "private use tags with skos:prefLabel" option available...)
>>>
>>> Ah, we hadn't spotted that.  Given that a vocabulary and its 
>>> notations are defined in a PDF document http://foo/bar.pdf (which is 
>>> distinct from the current document), I suppose I can refer to this 
>>> as follows:
>>>
>>> <#concept>
>>>  skos:notation "1.2.3"^^<#notation> .
>>> <#notation>
>>>  dc:description "The notation is defined in the document 
>>> http://foo/bar.pdf" .
>>>
>>> Would that be correct?  It appears to be consistent with the text in 
>>> SKOS Reference section 6.5.
>>>
>>> In this context, I can see little benefit in creating an XSchema 
>>> datatype, and requiring something like that would increase the 
>>> complication of the Recommendation we're writing.
>>
>> I would be rather sceptical about defining an explicit XML datatype, 
>> too.
>> Honnestly I've always been sceptical about the skos:notation stuff, 
>> because it seemed to be forcing people to make too complex things.
>> But in fact if your suggestion is correct (and it seems to be!) that 
>> would give a good balance, anchoring a string to a well-identified 
>> and accessible notation-defining space and still not defining this 
>> space formally. I'd be tempted to mention more explicitly this 
>> practice in the Primer, also (in which case it would be our turn to 
>> be thankful to you ;-)
>>
>> Now, I'm not a real expert in RDF datatypes, it would be great if 
>> someone on the list could validate your approach with more certainty. 
>> Especially, if it is something that shall be encouraged or 
>> discouraged (my two cents is that it should be encouraged if we want 
>> people to use skos:notation). Jeremy's opinion would be useful, as he 
>> co-authored [3]!
>>
>> Cheers,
>>
>> Antoine
>>
>> [3] http://www.w3.org/TR/2006/NOTE-swbp-xsch-datatypes-20060314
>>
>>
>>
>>>
>>> Thanks again for the comments.  Best wishes,
>>>
>>> Norman
>>>
>>>
>>
>>
>>
>

Received on Thursday, 9 October 2008 20:38:14 UTC