I’m often working on OpenStreetMap and Wikidata in tandem. With Wikidata typically distinguishing a municipality and its principal place (municipality seat or “capital”), even if they go by the same name, I sometimes come across towns or villages in OSM which are linked to WD IDs and corresponding Wikipedia articles which are not on the town or village, but on the municipality. To resolve this and establish proper mutual linking, I used to create a separate OSM node with place=municipality, move the wikidata- and wikipedia tags there and set the correct wikidata tag for the town or village.
is that this is the correct way of mapping a municipality. Via Overpass query, I am also able to establish that e.g. in southern Germany there are a number of municipalities not edited by me, which use a node with place=municipality.
From the English wiki page, I very recently learned that the municipality node should be included in the municipality’s boundary relation with the role “label”. I updated all of my edits accordingly, also setting role “admin_centre” to the village, where it was missing.
By occasionally creating municipality nodes in places where OSM village and WD/WP municipality are mixed, it happened that I touched three municipalities that are located in the small Austrian federal state of Vorarlberg. Based on my 3rd respective Changeset: 160922812 | OpenStreetMap, two guys now started bothering me. They claim that I am not entitled to create this type of nodes in Vorarlberg. Their line of argument is that other municipalities in Vorarlberg do not use them to present and that I have to align using them on the forum first.
Therefore, my two questions to the forum:
Do I completely misunderstand the concept of the municipality node? As said before, my understanding is that it defines the place on the map, where renderers should put a municipality label. From the German wiki page, I also understand that it is the place where wikidata and wikidata tags on the municipality should go to.
Is there a rule or an authority in OpenStreetMap that defines that certain documented ways of mapping/tagging shall not be used in a specific state of a country? So far, my understanding of OSM was that it is a worldwide map that is never complete, but is continuously improved on a best-effort basis. And that everybody is allowed to make meaningful contributions.
I am looking forward to your feedback and guidance.
A glance at overpass suggests that usage is hugely variable by country worldwide. There is exactly one in the UK, so I wouldn’t read too much into “British English tagging”
I’d suggest that the German page (one of the two countries where it is widely used) is a better option.
Edit: A look at an area of the other country which seems to have a lot of coverage (Sweden) finds data that seems partial. I’m guessing that admin tagging is complete, but this place tagging is just "filling in the gaps " of other places.
Place a node in the center of the municipality's boundaries (but not in the village) and add the following:
- place=municipality
- name=*
- wikidata=*
- wikipedia=*
Add point to boundary relation with role "label".
Or, if there is a town with the same name, add it to boundary relation with role "admin_centre".
My understanding of the English article of the boundary relation is that the same principles apply, but not going into detail how to tag the municipality node.
Every country is organized slightly differently; the relationship between place=* values and real-world places and administrative structures inherently varies around the world. Regardless of the country, it’s important to distinguish between roles: the label role basically says, “This is a different representation of the same thing,” whereas the admin_centre role says, “This is its administrative center.” In general, an administrative boundary and its label member usually do have matching wikidata=* tags in practice, except when Wikidata has a separate item about the populated place that’s represented only by the place point.
Here in the U.S., some mappers have used place=municipality points to represent the geographic centroids of local boundaries of “towns” in New York and Wisconsin and “townships” of some other states, because these boundaries completely partition their surrounding counties, unlike most local boundaries that are formed around a populated place. Confusingly, this is the antonym of a “municipality” in official and colloquial speech: outside of OSM, this term refers to an incorporated city or town that normally does not completely partition the surrounding county.
However, the local community is increasingly of the opinion that we don’t really need to map the centroid of a boundary as a place point, because a renderer should be able to generate the centroid automatically. (Currently, OSM Carto does, while OpenMapTiles does not.) If we don’t map the centroid, then there’s no label member, and there’s really nowhere for the place=municipality tag to go. This is fine, because the tag doesn’t really add much meaning anyways compared to other keys like admin_level=* and border_type=*. Meanwhile, the wiki tags would go on the boundary relation.
We would still map the place point when it represents a populated place at a non-arbitrary location. A good rule of thumb is that the place point represents the populated place if the place existed before it had an official boundary. But again, this is very dependent on the country.
In the present case, we are talking about sparsely populated rural areas with a lot of mountainside. So the areal coverage of the villages, which are the municipality seats, are much smaller than the municipalities.
(Please forgive me since I don’t understand German; the machine translation of the changeset discussion may have lost some nuance.)
You’ve modeled Bezau as a boundary relation with an admin_centre node and label node. The label is tagged place=municipality and has the same Wikidata tag as the boundary. As far as I can tell, this is accurate, but there may be an alternative approach that’s just as accurate. After all, a centroid point is really nothing more than a shortcut for a data consumer to avoid processing the whole boundary geometry. If you delete the place=municipality point, then the boundary relation should certainly retain its wikidata=* tag.
Just as in the U.S., the mappers who commented on your changeset disfavor the comprehensive mapping of place=municipality centroid points. The Austrian community is trying to be nuanced, allowing such points only where there isn’t already an administrative center by the same name. Presumably this would avoid redundant labels on rendered maps. By contrast, the U.S. community would discourage any centroid points, starting with townships (place=municipality) and counties (place=county) for now, but possibly extending to states (place=state) and the country node later on once enough data consumers can cope with the change. The Americans can’t really take the same nuanced approach as the Austrians, because the result would be quite incoherent in some states, due to local real-world naming practices.
Not exactly. I came across the municipality of Bezau, which was already modeled with the boundary relation Bezau (I guess, all municipalities are modeled with boundary relations in this area). However, the municipality’s main village Bezau had a wikipedia=* and wikidata=* tag assigned, which does not refer to the village, but to the muncipality. Since there also is a Wikidata Item on the village, I created a node for the municipality, tagging it with place=municipality, moving both tags over and setting the correct wikidata=* tag on the village node.
I did all of this following the guidelines on the OpenStreetMap Wiki.
It’s worth noting that worldwide, place=municipality is used on nearly 13000 relations and few than 3500 nodes. Indeed in Slovenia (which provides the map illustrating that page) it is used only on boundaries. So any data user wanting to interpret the relevant tags globally would need to look at the boundary relation. Clearly nodes are not the main carriers of this information on a worldwise basis.
My German is not good enough to be sure if the DE page is recommending tagging the label node additionally or instead of the relation. In any cases the reference to Germany in the first line of that page suggests the page may have been written specifically with Germany in mind. I would be cautious about assuming it applies to any other country.
As suggested in the changeset comments, a question in the Austria forum might be helpful.
But why should it be done differently in Austria? My understanding is that OpenStreetMap is a worldwide map. So if something applies to Germany, which is not country specific like traffic sign codes, it should apply to Austria and other countries as well. Also, there is no separate OpenStreetMap Wiki for Austria since they also speak German.
Maybe the question is “why should it be done differently in Germany”? From the examples I’ve seen, Germany may be the outlier.
I’d say administrative subdivisions are fairly country-specific in practice, and I wouldn’t expect any global wiki page to capture the nuances of mapping in any specific country. The concept of a municipality is certainly not global, and the English description “single urban administrative division having corporate status” is probably meaningless in most countries. Even as a native English speaker I’m not sure what it means - “corporate status” sounds more like something to do with companies than cities to my ears.
It’s also possible that it is entirely accurate for Germany (meaning that Germany mappers really do map this way). And it may well be accurate for other countries too - I didn’t do a systematic check on how those 3500 place=municipality notes are distributed globally. Note that even in Austria these nodes are used in some limited circumstances, so a bare “do not use” would not be correct.
(Also, did you mean “do not use” on nodes specifically? I don’t think there is any issue with place=municipality on boundaries).
Let me really just ask out of understanding: Why does it need two points on the map that have the same name and the purpose of painting the name of the municipality on the map?
I’m really just so confused about it because I’ve simply never seen this approach before anywhere where I mapped before. (And I don’t think I’m really that new to this project.)
Overpass Turbo of all admin-8 boundaries in Vorarlberg, Austria: overpass turbo
My understanding is that the town hall or similar has been used as the admin_centre (Example) and the subjective centre of the village as the label (Example) - but only very rarely!
By far the majority of these relations in this order (level 8) simply have a single node with admin_centre, which simultaneously accommodates the place=village|town|hamlet. I have never seen the concept with municipality before.
Referenced on the overview page of place: https://wiki.openstreetmap.org/wiki/DE:Key:place#Ortsangaben
A place=municipality is a ‘local authority’ (Gebietskörperschaft), i.e. when a village has joined forces with other villages for administrative purposes. But that is really not the case here.
I tried to machine translate the one Wikipedia article linked to this wikidata and I have a gut feeling that the whole article is complete nonsense – is that possible?
As far as I understand it, the Wikipedia article is about exactly the same thing as the main article, isn’t it?
What is the difference of the two wikidata items?
The wiki says you should first add a place with place=municipality and then add this to the relation as a label. But I’m afraid that the German wiki is not weill maintained / up to date.
It seems that the Cebuano Wikipedia was created by a bot based on GeoNames and other databases. They certainly do not know much about Germany and Austria. Therefore, you are right.
However, it makes pretty much sense to distinguish the municipality and its main village even if they have the same name. They are not the same entities. Compared to the other settlements in a municipality which have distinguished names, this would “shadow” the main village otherwise.
This page shows additionally where municipality comes from: More from the admin_border=7 than from the German translation of municipality. It may be the same problem that was discussed in Admin_level, boundary naming on OSM.org and translation
this is the only relevant aspect why it should be used at all and
given that the boundary relation can do the complete job of encapsulating all information on the municipality and
renderers rely on the boundary relation anyway
I wonder why we would ever need a node with place=municipality. What does it help with municipalities where villages were put together for administrative purposes?
Bodman and Ludwigshafen are two municipalities - that are actually quite far apart - but both have merged their administrative centres.
This is where the OSM wiki makes sense, because this is a “Gebietskörperschaft”.
Honestly, in my understanding of English, or at least Canadian English (which might be different from OSM English!), these are two towns or settlements that are one municipality due to some organizational or political reasons. So using one place=municipality on this node makes sense to me. Maybe the node could be the label of Relation: Bodman-Ludwigshafen (2784817) | OpenStreetMap.
