purl, vocabularies and ivoa.net

[nicolas moreau]

Dear all,

We thought about the uri based access of our vocabulary as uri being
identifiers (ids) for the vocabulary's words.

The representation (html, xml, json, ...) of these words should be as 
independant
as possible from the identification part.

That is why we thought about reusing what existed with `purl`.
`purl` doesn't carry much semantic about the representation in it, while
`www` does (in our opinions, since it is html oriented).

Thus, the model we proposed in the previous message.

The representation part could be added by some REST based mechanism,
like `?format=json`

A url like http://www.ivoa.net/rdf/Vocabularies/ is a good
place to store the RDF file that contains the concepts but not to create ids
for the concepts inside it.

As we said in the previous message, the purl pattern has our preference 
but as a second choice
  www.ivoa.net/vocab/theory could also be used as it does not carry any 
representational information.

Best regards,
     Nicolas


Le 25/05/2017 à 21:08, Patrick Dowler a écrit :
> There are existing urls that are published in recommendations, at least:
>
> http://www.ivoa.net/rdf/datalink/
>
> and there is an existing
>
> http://www.ivoa.net/rdf/Vocabularies/
>
> which appears to have a set of vocabularies (IAU and IVOA thesauri,
> UCD list, etc) that get maintained and released together (given the
> timestamps in the directory names).
>
> In that context, I think that something like
> http://www.ivoa.net/rdf/theory/ makes sense in the long term. The
> there are really two other issues:
>
> Where is the contained actually maintained and hosted? If it is
> maintained at another site then a redirect would work. If not, then
> a/the document submission process would have to support vocabulary
> updates and that would be used. We need that update process anyway.
> Supporting externally maintained/managed content under the
> www.ivoa.net namespace is a policy decision for TCG or exec. In my
> opinion, I'd prefer that the vocabulary maintainence process was good
> enough that this was not needed.
>
> Then there is the canonical identifier for the vocablary. Introducing
> purl.ivoa.net indicates to me that this identifier will be changed (in
> clients and docs) and I don't really see what that alternative
> hostname adds... why not just use http://www.ivoa.net/rdf/theory/ as
> the canonical identifier? This was the recommended style (from Norman
> Gray and Sematics WG) back when we spun the datalink vocabulary out of
> the spec and I don't think anything has changed. Does "purl" convey
> something special that really matters - more persistent that
> www.ivoa.net? Are we really trying to make a quick-and-dirty purl
> server with apache redirects?
>
> So, I would recommend not creating purl.ivoa.net and just using
> www.ivoa.net (/rdf/theory)
>
> Pat
>
> On 25 May 2017 at 02:15, Walter Landry <wlandry at caltech.edu> wrote:
>> Hi Everyone,
>>
>> For what it is worth, I would prefer to spell out things like this (no
>> one is going to know what 'purl' means).  I also do not like the use
>> of domains.  So I would go for
>>
>>    ivoa.net/permanent/vocab/
>>
>> Cheers,
>> Walter Landry
>>
>> Marco Molinaro <molinaro at oats.inaf.it> wrote:
>>> Dear Nicolas, dear all,
>>>
>>> so, the preferred solution would be along the lines of
>>>
>>> purl.ivoa.net mapped in DNS on the same machine as www.ivoa.net
>>>
>>> purl.ivoa.net/vocab/ to host vocabularies
>>>
>>> purl.ivoa.net/vocab/theory to redirect to Paris server
>>>
>>> Still, I think there's a problem with the
>>>
>>> www.ivoa.net/rdf/Vocabularies
>>>
>>> because we'll have or duplication of locations for vocabularies or to
>>> change the current behaviour of the endpoints.
>>>
>>> I wonder if this thing has to move out of the Theory/Semantics list to
>>> TCG level to make then a suggestion to Exec, because it's actually
>>> Exec that has rights to manage the ivoa.net domain, and this looks
>>> like a change in what's currently in there.
>>>
>>> Any thoughts on this?
>>>
>>> Marco
>>>
>>> 2017-05-24 14:18 GMT+02:00 nicolas moreau <nicolas.moreau at obspm.fr>:
>>>> Dear Marco,
>>>>
>>>> In order to simplify the redirection process if vocabularies are distributed
>>>> in several places,
>>>> we suggest to use the following pattern :
>>>>
>>>> purl.ivoa.net/vocab/theory/ConceptName
>>>>
>>>> and redirect it to Paris pages.
>>>>
>>>> The use of a purl prefix makes it explicit that we are dealing with a
>>>> persistent identifier
>>>> and not a simple url.
>>>>
>>>> www.ivoa.net/vocab/theory is acceptable too, it is simply not as clear if we
>>>> are
>>>> requesting a web page or a concept.
>>>>
>>>> Cheers,
>>>>      Nicolas
>>>>
>>>>
>>>>
>>>> Le 19/05/2017 à 08:42, Marco Molinaro a écrit :
>>>>> Dear Semantics,
>>>>> Dear Theory,
>>>>> (IVOADoc in CC)
>>>>>
>>>>> during the session this Tuesday, in the "SKOS in VO THeory" talk, it
>>>>> was proposed a change for the vocabulary for Theory VO with patterns
>>>>> pointing to
>>>>>
>>>>> http://purl.ivoa.net/ConceptName
>>>>>
>>>>> Now (apart from DNS aspects that can be managed on the DNS+IVOADoc
>>>>> side - that's why IVOADoc is in CC), ivoa.net has already a branch
>>>>>
>>>>> http://www.ivoa.net/rdf/
>>>>>
>>>>> that is meant to store vocabularies (Datalink one is already there).
>>>>> Also, the theory vocabulary will actually physically be in Paris, i.e.
>>>>> purl.ivoa.net will simply redirect to it.
>>>>>
>>>>> This email is to try to understand exactly how we can make this things
>>>>> work together, namely:
>>>>> - the purl.ivoa.net and its redirection to Paris (but possibly it
>>>>> should survive in ivoa.net and be used by other parties)
>>>>> - the relationship between purl.ivoa.net and www.ivoa.net/rdf
>>>>>
>>>>> Ideas, comments, .... are welcome, so that we can also start thinking
>>>>> of DNS changes in Trieste (hosting the IVOA.net domain and related
>>>>> DNS) and maybe in IUCAA (where the web part of the ivoa.net is
>>>>> physically located.
>>>>> I can take care of connecting with IVOADoc and ivoa.net sysadmins for
>>>>> the technical part.
>>>>>
>>>>> Cheers,
>>>>>       Marco
>>>>
>
>

-- 
-------------------------------------------------------------------------------------------------------------------
Nicolas Moreau,  Observatoire de Paris-Meudon-Nancay
LERMA 01 45 07 79 76 mail:nicolas.moreau at obspm.fr
-------------------------------------------------------------------------------------------------------------------