Open bhousel opened 6 years ago
Using wikipedia
and wikidata
instead of (or in addition to) brand:wikidata
and brand:wikipedia
is also a popular issue.
For example wikipedia=en:McDonald's
instead of brand:wikipedia=en:McDonald's
.
EDIT: I opened issue for that in iD - see https://github.com/openstreetmap/iD/issues/6416
Good point @matkoniecz - let's add it in addition to..
Is somebody interested in processing should list (making edits in their local area or making remote edits based on proper research without blind edits)?
I made a simple generator after I noticed some suspicious matches that seemed to contradict what I mapped.
Example output, part of output for {'lat': 51.5148, 'lon': -0.1018, 'max_distance_in_degrees': 0.05, 'name': "Central London"}
Sainsbury's Local shop=convenience, (tagged shop=supermarket)
https://www.openstreetmap.org/way/483801992
Disney Store shop=gift, (tagged shop=toys)
https://www.openstreetmap.org/node/2379057654
Jack Wolfskin shop=outdoor, (tagged shop=clothes)
https://www.openstreetmap.org/way/186337067
Argos shop=catalogue, (tagged shop=department_store)
https://www.openstreetmap.org/way/264127746
The Body Shop shop=cosmetics, (tagged shop=beauty)
https://www.openstreetmap.org/way/183360808
The Body Shop shop=cosmetics, (tagged shop=chemist)
https://www.openstreetmap.org/node/3357929205
Greggs shop=bakery, (tagged amenity=fast_food)
https://www.openstreetmap.org/node/748091223
Greggs shop=bakery, (tagged amenity=fast_food)
https://www.openstreetmap.org/way/654507605
Vans shop=shoes, (tagged shop=clothes)
https://www.openstreetmap.org/node/3687970999
Vans shop=shoes, (tagged shop=clothes)
https://www.openstreetmap.org/way/273610366
Claire's shop=jewelry, (tagged shop=clothes)
https://www.openstreetmap.org/way/159498892
If someone is interested I may generate and publish data for their location - it would be especially useful in places known to be well mapped as it may reveal some overzealous matches.
Tagging @mvexel here - he was interested in adding some challenges like this to Maproulette, now that it can do quickfixes.
I've been slowly adding these to maproulette so they can be cleaned up. Usually about 400-500 a week are getting cleaned up.
I would explicitly note in maproulette tasks that proposed changes are not always going to be correct and that mistaken suggestions should be reported, sometimes NSI needs to be fixed - not OSM data.
For example https://maproulette.org/browse/challenges/9930 is phrased like that this change is guaranteed to be right. In such case one should run rather a bot edit.
Task along these lines:
"brand:wikidata"="Q3069980"
matches previous entries for Monsoon and Accessorize (clothing shops).
These have been updated to have individual wikidata items, but the old tagging is already present in OSM. In theory it's safe to just update based on the name/brand tags since the locations are explicitly tagged by mappers as part of the overall business.
@frodrigo and @jocelynj, would Osmose be able to run checks from the name suggestion index (NSI) against the OSM planet, perhaps daily or weekly, and make the resulting lists of features that need fixing available through an API?
@frodrigo and @jocelynj, would Osmose be able to run checks from the name suggestion index (NSI) against the OSM planet, perhaps daily or weekly, and make the resulting lists of features that need fixing available through an API?
Maybe all of this is already done https://osmose.openstreetmap.fr/en/map/#zoom=10&lat=47.1859&lon=29.071&item=3130&level=1%2C2%2C3
The API is also available, but the doc page in temporary offline. https://wiki.openstreetmap.org/wiki/Osmose/api/0.3
But the API content can be from the links on this page https://osmose.openstreetmap.fr/en/issues/open?item=3130
cc @nlehuby
@bhousel To display a dashboard (#6496), in which format would you like to get the data? For example, in the brand list on nsi.guide, it might be useful to show the number of unresolved problems next to each brand. It looks like this number could be gathered from Osmose, since they already run the NSI checks against the OSM planet. Once you tell what exact format you need, perhaps the file could be produced and get served directly by the Osmose infrastructure, or else (in case Osmose would rather not have this inside their codebase) perhaps we can set up a daily cronjob in the Wikimedia cloud which would call the Osmose API to produce the output file. Hence the question: What format/schema to use?
While we're processing the planet file, we can also produce some output about which features need updating (by comparing against the accepted entries in
config/canonical.json
).e.g.
brand:wikidata
,brand
, etc)amenity=restaurant
should beamenity=fast_food
) maybe more