Page MenuHomePhabricator

Auto-convert "wikipedia"→"wikidata" tags for geoshapes service
Closed, DeclinedPublic

Description

Given Wikidata ID, Kartotherian geoshapes service should find the right shape even if it is only tagged with the Wikipedia tag, without Wikidata ID.

We could add an extra post-download step, that would go through all rows, find any that are missing Wikidata IDs, and look them up in Wikidata. Ideally we would simply import the whole Wikidata ID database, or use their API to resolve a large set of Wikipedia article names to corresponding Wikidata IDs, and add them to Postgres.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript
Yurik lowered the priority of this task from Medium to Low.Nov 26 2016, 4:41 AM

Still pondering if this is a good idea, or if we should simply use the OSM's Wikidata ID, and encourage its creation and cleanup.

Still pondering if this is a good idea, or if we should simply use the OSM's Wikidata ID, and encourage its creation and cleanup.

I would love to do both. We can do it, and I think we can publish the list of OSM objects that have a Wikipedia ID, no Wikidata ID, with the suggested/auto-filled Wikidata ID. A list we should publish to the community and track the size.

I would love to do both. We can do it, and I think we can publish the list of OSM objects that have a Wikipedia ID, no Wikidata ID, with the suggested/auto-filled Wikidata ID. A list we should publish to the community and track the size.

I wish it was that straightforward :( For example, in UK they have "ceremonial" and "administrative" parishes. Ceremonial describes the whole area, whereas administrative usually describes the same area minus the city at the center of it. When the city grows too big, it becomes its own administrative region. Wikipedia articles, and by extension, Wikidata IDs describe both at once, but OSM would usually have two distinct objects - one with and one without the center city. Wikidata sometimes has a separate item for the admin parish, but it is not well organized, and has many errors. Simply drawing based on Wikipedia link would cause two different areas to be drawn, partially overlapping one another.

debt subscribed.

Moving off the sprint board - the Discovery team won't be able to finish this work at this time.

Given the questions above and the lack of interest in it, I'm going to close it. If someone wants it, they can re-open, although it's unlikely to be worked on at the WMF.