Re: [SKOS] Re: ISSUE-224 [was: Agenda - 2009-07-07 SWD telecon]

Hi Sean, Ralph,

Trying to clarify my previous mail, and react to your discussion

1. I agree that the only problem (syntactic conformance) lies with with example 22, in the triple:
<Potassium> skos:prefLabel "K"@en-x-chemicalsymbol .

2. In my mail I wanted to explain that I could live with the following practice: for the *particular situation* of representing notations with prefLabel (or altLabel) then there could be no language tag (I hope this answers your question, Sean). The triple in example 22 may thus become:
<Potassium> skos:prefLabel "K" .

Yet I was uncomfortable with that solution, because it does not allow easily to represent the example in the Primer [1]
iso3166:FR skos:prefLabel "France"@en ;
  skos:prefLabel "FRA"@en-x-notation-threeletter ;
  skos:prefLabel "250"@x-notation-numerical.

Indeed, if we opt for a "tag-less" representation of notations, we should have
iso3166:FR skos:prefLabel "FRA" ; skos:prefLabel "250".
which is inconsistent.

*But* in fact there could be a way out: namely, we can defend that a notation like "FRA" is language-dependant (it's the abbreviation of an English word) and keep something like:
iso3166:FR skos:prefLabel "FRA"@en-x-not3 ; skos:prefLabel "250".
This is consistent, may be less frown upon by the i18n WG, and still allows to represent the example in a relatively satisfying way.

To end with the consequences of this wrt. the resolution that was made at the teleconference yesterday [2]: I'll happily live with the current solution, which is the best and simplest. But *I can also live* with a solution that would:
- keep to (a)
- complete (b) as in:
"note that such use of private subtags to transmit data unrelated to language or language choice may violate BCP 47. To alleviate this, implementers may opt for representing notation literals without language tags."
- add in 6.5.4 a new example with the triple: <Potassium> skos:prefLabel "K" .




> At 03:49 PM 7/7/2009 +0100, Sean Bechhofer wrote:
>> On 7 Jul 2009, at 14:15, Antoine Isaac wrote:
> ...
>>>  2009-06-29. Addison Phillips on I18N issue
>>>>  --
>>> Re. this issue, I can live with the option where we would recommend  
>>> to use pref and altLabels with literals that have no language tag.  
>>> Even though this is clearly less flexible (e.g. not being compatible  
>>> with the country code example).
>> As I understand the comment, the problem is specifically with example  
>> 22 rather than the use of language tags with pref and altLabel,
> That is the way I read the I18N Core WG comment as well.
> It should not be too difficult (nor require restarting Proposed Rec
> review) to alter the example, right?
>>> I also volunteer to start a discussion with the i18n people on  
>>> whether the use of private use tags is merely frowned upon, or if  
>>> they would still formally object if our examples were just made  
>>> compatible with the syntax for private use tags.
>> That would be great -- thanks Antoine.
> I'm nervous about asking for approval :)
> They are very careful about their prose.  They specifically wrote
> "frowned upon" and not some stronger language and that's
> our loophole.  I believe this could be related to discussion we
> had at our May 2008 face-to-face [1] as well.
> I would recommend instead that we offer to (a) modify the
> example(s) to be syntactically conformant and (b) offer to add
> a phrase such as "note that such use of private subtags to
> transmit data unrelated to language or language choice may
> violate BCP 47".
>>> I indeed understand that syntactially wrong tags are to be avoided  
>>> in the rec. But if we make explicit that we're really not expecting  
>>> the use of private tags to be common, then maybe the i18n would be  
>>> happier on this specific point (which is different from the  
>>> syntactic validity, again).
> I don't think we're going to get them to admit that they're "happier"
> with us documenting a practice upon which they "frown" :)
> [1]

Received on Wednesday, 8 July 2009 16:39:17 UTC