Discussion at the task force
Exclusion of villages, neighboorhoods
Since most villages (and neighboorhoods) are not an administrative unit, how can we 'list' them on for example a municipality item if we can not use P150? Michiel1972 (talk) 09:18, 4 October 2013 (UTC)
- Is it correct to use has part (P527) as replacement of contains administrative territorial entity (P150) for non-administrative localities, e.g. see my switch from P150 to P527 here ? Michiel1972 (talk) 09:24, 4 October 2013 (UTC)
some administrative divisions contain only a fraction for some of its "subdivisions". This property also assumes the reverse property is true, i.e. the subdivision is fully part of the parent division, but here also it may have several parent divisions.
Common examples include
- French cantons (their subdivisions are not always full communes but only "fractions" of communes)
- Belgian linguistic communities (région de Bruxelles belongs to several ones)
- legacy subdivisions of Spain grouping communes belonging to several provinces (provinces not covering these groups completely).
Do we need to create separate entities for these fractions, or shouldn't we create another property than this one when the listed subdivision(s) is/are not fully included ? (in that case the reverse property could also be created for partial parentship, unless we assume that if there are several parent divisions of the same subdivision type, the child subdivision is split between these parents).
Note also that subdivisions of a parent subdivision are not always the same type. Example in France:
- departments are subdivided into arrondissements, these arrondissements were subdivided in parallel into cantons, and in communes (communes may be fully part of a canton, or a canton may be fully part of a commune, or a canton may cover several communes and a part of some other communes)
- starting March 2015, news (larger) cantons will no longer subdivide arrondissements but will subdivide departements, but they will still cover full communes or fractions of communes.
- groupments of communes (EPCI's such as communautés de communes or "syndicats" or the new "métropoles" starting by Lyon) have a seat in a well defined commune (so in a well defined canton, arrondissement, departement, region for that seat) and only include full communes; but these groupments may cover communes belonging to several cantons, or arrondissements, or departements, or régions. An EPCI with its own fiscality only group communes, not the other regional entities, but other "mixed" groupments (the "syndicats" like SIVU and SIVOM or the "pays", or syndicates managing regional parks) may have other entities within their members, such as the departemental or regional council, or chambers of commerce, or portuary authorities, or basin agencies for water management.
- Starting in 2015, the creation of métropoles creates a new separate kind of subdivisions of departements : the "métropole" (a new kind of EPCI, with strong power, to which belong full communes, and over which the departemental council is no longer competent), and the remaining communes of the departement (they may only be members of other kinds of EPCIs but not the métropole, and they are still coordinated by the departemental council).
- It is even possible that regional councils will no longer be competent in new "métropoles" (but for now regions are being reformed and some departements may change regions), so that métropoles cannot be considered as subdivisions of regions (for the regional council) even if the state-level administrative system (of prefectures/subprefectures) do not change their area immediately (it is envisioned, for the state-level, to abandon the distinction between regions and departements and completely renew itself within a new landscape with either métropoles or groups of EPCIs (prefectures/subprefectures will be relocated in them or will adapt to their autonomous evolution).
In summary, we need clarification about the usage of this property, that has been used to list sub-entities that are not always fully included in their scope. I suggest a separate property for partial coverage.
The alternative being to no longer list in this existing properties any entity that is partly covered, but only "ad hoc" entities created to mean they are only a common fraction of some other entitities (e.g. for the intersection of a commune and a canton in France), but in that case these "fractions" are NOT really administrative subdivisions by themselves (they cannot be a commune, or a canton, but a new specific subclass of administrative subdivisions).
Final note: there DOES exist administrative entities encompassing territories in several countries, generally for cross-border cooperation, or for territories under joint management and specific local legal systems (such as the international airport of Bâle-Mulhouse, both fully built in France, or the CERN facility built in France and Switzerland, or some maritime areas like the Gulf of Biscay shared between France and Spain, or a small island in the Pyrénées which becomes French or Spanish every 6 months, or transportation systems for commuters such as between France and Luxembourg, or France and Switzerland) ! These are working under regulation of a joint authority (under the status of "European Interest Group", with specific fiscality shared between member countries)
We really need another property for former subdivisions. Currently there's a lot of confusion caused by the callssification of some geographical entities in their current or former administrative entities and Wikidata simply gives false data when we recurse through them.
An obvious example comes with French cantons, that have radically changed in terms of structure: before march 2015, they were administrative subdivisions (fourth level) of arrondissements (third level). Since March 2015 they are no longer administrative divisions and are actually electoral subdivisions (no level) of departments, and no longer of arrondissements. Many articles in various wikipedia editions are incorrectly sorting them.
The situation exists in fact in many countries that have territorial reforms : we would like that this propery refers only to current subdivisions to make sure every location is correctly sorted with a complete and coherent way. For that we should have another property for all other former subdivisions (which are frequently incomplete but for which there may exist data related to articles in Wikipedia (and elsewhere) discussing the history of these geographical entities and locations.
In other words: please add a distinct property for former subdivisions. The alternative is to use this property but with a separate "category" for former subdivisions.
Note: a former administrative subdivision may still have a current usage for something else even if it is no longer administrative. This is the case of former French cantons (before march 2015) which are no longer administrative subdivisions of arrondissements (whic themselves are still administrative subdivisions of French departments) but are still electoral subdivisions of French legislative constituencies (which are themselves still electoral subdivisions of departments).
In general there are lot of errors now with the classificiatons of French communes in various Wikipedias, or in Commons and elsewhere. Those communes should NO LONGER be listed as administrative subdivisions of any canton (today's cantons are electoral only) but only as administrative subdivisions of their arrondissement. But even before March 2015, not all communes were subdivisions of former cantons (this is wrong for all comunes that were split across multiple former cantons, and the situation remains for today's cantons: for now Wikidata list those communes as being "administrative subdivsions" of all these cantons, but it is wrong: the real subdivisions of cantons are either communes included entirely, or "fractions cantonales" (where are both "electoral subdivisions" of cantons, and "electoral subdivisions" of communes).
The data model need clear adjustments. And at least the usage of "administrative subdivision" for today's French cantons is completely wrong. So in summary we need to separate the property "administrative subdivisions" from:
- "former administrative subdivisions" (e.g. all former French cantons before March 2015, but NONE of today's French cantons; other examples: all former regions of Morocco before 2015)
- "electoral subdivisions" (e.g. all today's French cantons, all today's French legislative constituency, all today's senatorial constituencies)
- "former electoral subdivisions" (e.g. older French cantons, older French legislative constituencies, or many old European constituencies in UK)
There are many examples in many countries.
For now I request this new property:
|fr||anciennes subdivisions administratives||(liste des) anciennes subdivisions directes d'une entité territoriale administrative||comprenait|
était composé de
était divisé en
était subdivisé en
|en||former administrative subdivisions||(list of) former direct subdivisions of an administrative territorial entity||comprised|
was composed of
was divided into
was subdivided into
had administrative division
had local government areas
had rural cities
Data Type: Element
|subproperty of||has parts (P157) -- or possibly: a new similar property "had parts"|
When this new property is defined, we'll be able to sort the mixup in Commons and Wikipedia (articles and categories) and elsewhere by using it for all former subdivisions. Note that current subdivisions may be homnyms of former subdivisions but their composition will be different. In Wikipedia, we frequently have the same article or category speaking about both the former and the current subdivision.
Ideally however we need also the two other properties for non-administrative subdivisions (I gave the two examples of electoral subdivisions; some people may want to include as well postal subdivisions, or judiciary subdivisions, all possibly with "current" state or "former" state for articles about the history of territories).
- @Verdy p: We don't need "former" versions of every property. This is what end date qualifiers and normal ranks are for. --Yair rand (talk) 17:40, 9 January 2017 (UTC)
- Once again this is not at all about the date but the **nature** of the new cantons, that were subdivisions of arrondissements in the past, but not longer the case at all. And that biside their similar name as "canton" are not at all the same kinds of entities, administratively, and electorally, judiciarily. These are completely different entities (iondependantly of their geographic areas which are also completely different: there's absolutely NOTHING in common between the two kinds). They have the same name but we need some distinction, and they are also NOT at the same hierarchical level. The new "cantons" are in fact purely electoral for the departmental elections and have no use anywhere else. The judiciary map was NOT updated to use the new cantons, in fact the judiciary map is since long detached from cantons.
- You want to restrict only on a date difference, but this is a very tiny part of the differences, and both types continue to coexist in texts, but noàt for the same purpose. The new cantons should have been named "circonscriptions électorales départementales" and "cantons" left for the former entity type (no longer used for elections but still referenced and active in French law, notably in the Rural Code). Verdy p (talk) 18:08, 9 January 2017 (UTC)
Note: given that there is now a constraint on P131 that the things it connects should be "contemporary", ie both should have coexisted at some point in time, avoiding anachronism, I have proposed a new property "located in present-day administrative territorial entity", to relate historical administrative units to those in the present. Jheald (talk) 08:52, 21 February 2017 (UTC)
Deprecating rank with "end time" qualifier
Is it possible to add a validation rule to enforce "deprecated rank" on any entry that has an "end time" < "now" qualifier, as long as there are other P150 entries that do not have the same qualifier? This way United States of America (Q30) would not mix together the current states with the obsolete entries, but it would not trigger in case all of the entries have been deprecated? --Yurik (talk) 22:40, 27 January 2018 (UTC)
- I don’t think I agree that these statements should be deprecated at all – they’re not wrong, after all. If you care about truthy statements for queries I’d prefer making all the current statements preferred. Or is there a lot of existing convention to make these statements deprecated? —Galaktos (talk) 00:31, 28 January 2018 (UTC)