Use of cpsv:administrativeLevel, NUTS code

I don't see the added value from NUTS to CPOV. From a CPSV point-of-view, I can follow the logic to a certain point. Public services can often be organized according to territorial criteria. Only drawback for us is that NUTS remains on a fairly high statistical relevant level, while public services need more detail. But you could use as a starting point.

The organisations behind those public services, though, are in reality most often not organized according to NUTS-criteria. Either they demand more detail then NUTS-3 or they work for a group of NUTS-2 instances...

Another con is that NUTS is purely territory-based, while public services can be community-based. For us and in terms of CPOV, NUTS is rather a constraint in stead of an improvement.

The administrative, geographic and judicial structures (organizations) of teritorial units are not homomorphic.

To my knowledge NUTS was (initially) intended to serve as a resource for administrative organisation but ended up structured in terms of geographical & demographic criteria that are suitable rather for statistical purposes and fit well the needs of Eurostat. 

The need for an organization based on administrative critaria still remained. This was the reason why MDR, after several discussions with EuroStat, decided to create a NAL (Named Authoroty List) called Administrative Teritorial Units (soon to be published) that does similar work to NUTS but organised according to administrative criteria.

Thus one should use NUTS and Administrative Teritorial Units NAL based on the organization criteria: geographic or administrative

PS: MDR has also discovered that anotehr table should be created that provides an organization based on judicial criteria. 

PPS: If you wonder whether they can be mapped then I assume that some loose/approximate mappings could be traced between the two taxonomies but not exact ones as they are not homorphic.

To complete what Eugeniu has written already. The Publications Office of the EU is going to publish a named authority list (NAL) on 29/02/2016 in the Metadata Registry: http://publications.europa.eu/mdr/authority/index.html. The table will be called Administrative Territorial Units.

Initially we had envisaged to create one table aligning the different territorial units, but in discussions with the colleagues from Eurostat we realised that NUTS regions may have the same name as administrative units, but they do not necessarily represent the same territory. That is why we have abandoned this exercise and focused on creating a table with administrative units only. Work on a separate table with judiciary territorial units is in preparation.

Sources for this table have been the Committee of the Regions data and ISO 3166-2 subdivisions. The table only covers level 1 and 2.

We intend to use the following defininition to describe the table:

eng [Part 2 'country subdivision code') of ISO 3166 establishes a code that represents the name of the principal administrative divisions, or similar areas, of the countries.

Where possible we have tried to base the information on official sources. For example:

Ireland: Local Government Act, 2001 - Part 2, 10 (Irish Statute Book), updated in 2014 "The Regional Authorities were dissolved in 2014 and were replaced by Regional Assemblies, see SCHEDULE 1 " http://www.irishstatutebook.ie/eli/2014/si/573/made/en/pdf

France: http://www.legifrance.gouv.fr/affichTexteArticle.do?idArticle=LEGIARTI000019241099&cidTexte=LEGITEXT000006071194&dateTexte=vig et http://www.gouvernement.fr/action/la-reforme-territoriale

Italy: http://www.quirinale.it/qrnw/statico/costituzione/costituzione.htm#P2T5, Titolo V

Luxembourg: http://www.luxembourg.public.lu/fr/actualites/2011/12/16-communes/

As soon as we have an XML or RDF sample, we can post it here ...

Very interesting and a good possibility. Only remark is that the administrative units here may not be defined solely on territorial concepts.

Maybe I'm a bit prejudiced by the situation in Belgium. We separate community-based and region-based public services and public organizations. That means we have a Flemish Region and a Flemish Community that are not alike. Brussels eg. is it's own Region, falls stricto senso under the Flemish Community, but has separate administrative units.

So we can't rely solely on territory-based criteria to enrich public organizations. Administrative Territorial Units offers a correct way out on the premise that it can differ from NUTS (which I read in Willem's comments).

I prefer a MDR NAL to a typed literal - happy it will have that form.

All understood. The next version of the doc will show a relationship of dcterms:spatial linking an Organisation to Administrative Territorial Units (which IMHO should be defined as a subClass of dcterms:Location)

Just a clarification, the administrative level property of the CPSV does not refer to the a geographical area or location. It was included rather to indicate whether an specific public organisation is a local, regional, national or supra-national authority. 

The issue of the geographic area under the jurisdiction of a public organisation is a different discussion and hence needs to be modelled using a different property. 

In the meeting on 2016-03-09 we only had a short time to discuss this issue. I took an action to investigate how we might include geometries in territorial units. This is pretty straightforward - we can just use the Location Core Vocabulary's locn:geometry property which can link to a geometry, however it's encoded. My proposed text in the doc is below. I notice in doing this that the cardinality was set at [1..n]. I think 0..n is more appropriate.

Property: spatial [0..n]

This property links an organization to the administrative region(s) that it covers. The value of the properly should be the URI of the region as defined in an authoritative list of regions. In Europe, this is likely to be the Administrative Territorial Units[1] Named Authority List maintained by the Publications Office's Metadata Registry.

In the RDF release of the CPOV, this property is bound to dcterms:spatial.

The ATU list does not include a geometry. That is, the territory is only identified by its name not its spatial coordinates. This is likely to be the case for similar lists. If geometries are available for the Public Organization's territory, they can be linked from the territorial unit using the Location Core Vocabulary's locn:geometry property[2].

[1] http://publications.europa.eu/mdr/authority/atu/

[2] www.w3.org/ns/locn#locn:geometry

+1. The cardinality within scope of CPOV should indeed by 0..n.

ATU NAL is to be preferred.

Actually, we've just removed all cardinalities from the doc. Those are appropriate for an application profile, but not for a core vocab definition.

Proposed text accepted during the meeting on 2016-04-14