[Tagging] RFC: remove alphanumeric code visible in infoboxes at OSM Wiki linking to Wikidata

[Mateusz Konieczny]

Apr 3, 2022, 01:46 by dieterdreist at gmail.com:

>
>
> sent from a phone
>
>> On 3 Apr 2022, at 00:47, Mateusz Konieczny via Tagging <tagging at openstreetmap.org> wrote:
>>
>> But I tried to fix some structural issues on Wikidata
>> in past and it was not easy, and I am not planning to 
>> spend my time on it.
>>
>
>
> this is also my experience, it is very complicated to fix wikidata, first you have to find what is already there, often different items which are partly overlapping, and you would have to somehow merge or link them, or create meta items which link them. There is also the issue that you can only have one wikipedia article for every item, and IIRR you can also have the same wikipedia article only once in wikidata.
> For example in OpenStreetMap there is a distinction between place as a socio geographic object and administrative entities, in wikipedia both concepts usually or often are treated in the same article (this is less relevant for wikidata on tag pages, but demonstrates the kind of problems you run into, and is relevant when you want to link OpenStreetMap data to wikidata). Wikidata is easy when you deal with persons, e.g. find all women painters of the 19th century, because it is clear what is one person, but it is less clear when it comes to geographic and political entities, where the notion what they are or what is meant change and can be ambiguous. What is “Rome”, a city? A municipality? A province like area with a special status? Is the Vatican city part of it? Politically it is not currently, culturally it is. 
>
> Another example for the city realm: in wikidata, the city of westminster is an instance of a city, in OpenStreetMap it is not. What is your suggestion how this could be fixed?
>
> Maybe everything can be fixed, but from an OpenStreetMap perspective, wikidata is often a mess, specifically for geographic entities, and we are here to enjoy mapping in osm, not to loose our time in an uphill battle with wikidata bots. Adding “instance of” properties to wikidata objects has an infinite complexity, you could follow link after link, and reflect about the meaning, and maybe also refrain from adding the link, or not because  tomorrow some bot will come along and add it without hesitation anyway ;)
>
> I am all open to let people link wikidata and OpenStreetMap, but these are really just hints for loosely related concepts, and every measure to avoid the misconception that we are looking at the exact same thing is welcome, therefore I support Mateusz’ initiative to remove wikidata from the infoboxes, while retaining the link somewhere less prominent.
>
In general, it is much easier to fix blatantly wrong data in OSM than 
in Wikidata.

For further complexity (for people who can easily fix issues listed above):

- there is some structure ( https://www.wikidata.org/wiki/Q220 representing Rome,
of type "big city" and "border city" and "abolished municipality in Italy" among other,
last one is represented by "abolished municipality in Italy" at 
https://www.wikidata.org/wiki/Q3685476 )
I tried to use such structure, ended with plenty of trouble which I tried to monitor/control
with https://github.com/matkoniecz/wikibrain/blob/master/test_wikidata_structure.py

For example https://www.wikidata.org/wiki/Q3720557 (train line) was classified as
event, Maria columns were classified as events
( https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat&oldid=1359739358#How_to_prevent_Maria_column_from_being_classified_as_a_process? )
and they may still be (I stopped fixing such issues as people were restoring
classification that was classifying wooden churches as events and so on)

- there is Cebuano Wikipedia which is almost entirely generated automatically
by someone running bots (over 6 million articles).
Someone else run bot on Wikidata and created Wikidata entries for all of them
Noone bothered to ensure that this Wikidata entries are not duplicating already existing
ones.
Now editors are expected to deduplicate millions of entries.
Such approach to bots (almost exact opposite to OSM) is typical.
And while fixing individual duplicates is easy (with merge tool which can be enabled in
gadgets in preferences), just Cebuano bots create millions of them.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20220403/53025db5/attachment.htm>