W3C home > Mailing lists > Public > public-gld-comments@w3.org > April 2013

Re: dataset and Dataset

From: Phil Archer <phila@w3.org>
Date: Sat, 06 Apr 2013 10:21:22 +0100
Message-ID: <515FE912.8030104@w3.org>
To: Bill Roberts <bill@swirrl.com>
CC: RaphaŽl Troncy <raphael.troncy@eurecom.fr>, Makx Dekkers <makx@makxdekkers.com>, public-gld-comments@w3.org
*If* the WG decides to make this change, i.e. replace dcat:distribution 
with dcat:hasDistribution etc. then, after introducing 
dcat:hasDistribution, we have a choice:

1. delete dcat:distribution;
2. retain but deprecate dcat:distribution.

Arguments in favour of 1:
- it removes the danger of confusion;
- it reduces the chance of dcat:distribution still being is use in 5 
years' time.

Arguments in favour of 2:
- existing implementations that use dcat:distribution will be unaffected;
- we would avoid having to go through another round of last call.

Here's the rule book on Last Call [1]

"Ideally, after a Last Call announcement, a Working Group receives only 
indications of support for the document, with no proposals for 
substantive change. In practice, Last Call announcements generate 
comments that sometimes result in substantive changes to a document. A 
Working Group SHOULD NOT assume that it has finished its work by virtue 
of issuing a Last Call announcement."

In other words, if LC comments lead to changes that would cause an 
existing implementation to break, you need to do another LC cycle. You 
do that until the only comments received make no substantive change. 
*Then* you can go to CR - at which point it's about implementations, 
experience of which will either confirm the spec is fit for purpose and 
'works' or... send it back to Last Call.



[1] http://www.w3.org/2005/10/Process-20051014/tr#last-call

On 05/04/2013 17:46, Bill Roberts wrote:
> Fair enough - it seems from all the responses that this is not in fact a common practice.  Maybe I just think it is because some of the ontologies I use frequently (eg data cube) use it, and I use it in ontologies I create myself.
> I see the point about using more distinctly different predicates and classes and certainly wouldn't object to a change along those lines.
> Thanks for the discussion!
> Best regards
> Bill
> On 5 Apr 2013, at 17:13, RaphaŽl Troncy <raphael.troncy@eurecom.fr> wrote:
>> Dear all,
>>> I followed the idea of using a property to point to a class with the
>>> capitalisation of the class name as the only difference in some recent
>>> vocab work - and got push back.
>> I would also recommend to push back on this, and disagree with Bill, this is not an established practice (or at least, I would like to see evidence of the contrary), nor a practice to encourage.
>>> I was told by non-Sem Web data modellers
>>> that the more general convention is that properties (data type
>>> properties) should be nouns and relationships (object type properties)
>>> should be verbs.
>>> Things like org:hasSite cf. org:classification fit in with this.
>> Yes, and remember that we are talking about directed graph, and that a good practice is to give the direction of the property in its name, thus the hasXXX or the isXXXOf pattern.
>>> Based on that, if we were starting from scratch I'd argue for
>>> dcat:hasDataset or dcat:includesDataset but it may be too late now.
>> Why would it be too late? This is last Call. So this is the time. After, rec, this would be too late.
>> Best regards.
>>   RaphaŽl
>> --
>> RaphaŽl Troncy
>> EURECOM, Campus SophiaTech
>> Multimedia Communications Department
>> 450 route des Chappes, 06410 Biot, France.
>> e-mail: raphael.troncy@eurecom.fr & raphael.troncy@gmail.com
>> Tel: +33 (0)4 - 9300 8242
>> Fax: +33 (0)4 - 9000 8200
>> Web: http://www.eurecom.fr/~troncy/


Phil Archer
W3C eGovernment

+44 (0)7887 767755
Received on Saturday, 6 April 2013 09:21:48 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 10 October 2018 10:43:22 UTC