Forwarding to the Wikimedia Maps mailing list.
---------- Forwarded message ----------
From: Edward Betts <edward(a)4angle.com>
Date: Wed, Jun 7, 2017 at 11:39 PM
Subject: [Wikidata] Tool for users to add wikidata tags to OpenStreetMap
I've built a tool for mappers to match things in OSM with Wikidata and add
appropriate wikidata tag to OSM.
Users can search for a administrative area or city, then pick an area to
analyse. It works best if mappers pick areas they're familiar with.
The matcher can take a few minutes to run. It grabs items from Wikidata and
figures out a target set of tags and keys to search for. Then it downloads
data and looks for matches. The matching is based on tags and names,
the wikipedia tags aren't considered.
Map data is from the OSM Overpass API. Large (more than 1,000 sq km) or
areas might fail with a timeout error.
The results are cached in the system and are available here:
Once the matching process is complete the mapper is given a tabbed page with
the results. The five tabs are:
Match candidates - things on OSM that might be considered for tagging
Already tagged - matches that are already tagged in OSM
No match - items from Wikidata with no match found in OSM
Wikidata query - the query used to find Wikidata items
Overpass query - the Overpass query to find OSM objects in this area
To start tagging the mapper is able to login to OSM via OAuth. Tick boxes
appear next to the likely matches, the mapper can tick the box next to the
matches they want to upload, then add a change comment and upload them using
their own OSM account. Uploads within an area are combined into a single
If the mapper sees an obviously incorrect match they can use the 'report bad
match' option to warn other mappers and provide feedback that I can use to
improve the algorithm.
This tools doesn't add any new objects to OSM. The only change it makes is
adding a wikidata tag to existing things.
My approach is to aim for a one-to-one mapping between Wikidata and OSM. If
there are two or more things in OSM that look like a Wikidata item then it
isn't a good match. This means for example that most road and rail bridges
won't be tagged because they are represented as two OSM ways. I might change
this at some point.
There are occasional duplicates in Wikidata, this tool should spot them and
refuse to add wikidata tags until the Wikidata duplicate is resolved.
The bug/todo list is here: https://github.com/EdwardBetts/osm-wikidata/
Any ideas or suggestions are welcome.
Wikidata mailing list