Page MenuHomePhabricator

Update /page/definitions to use lexemes from Wikidata
Open, NormalPublic

Description

We should update the /page/definition endpoint to use lexemes from Wikidata ASAP, because the current results are:

  1. Very poor. They are blobs of arbitrary HTML from which clients are required to strip markup. In some cases, they even include messages between editors in the page text itself.
  2. Very expensive to create.
  3. Only available for the content in the English Wiktionary.
  4. Overly broad in scope. A little-known fact about the Wiktionary project is that each Wiktionary aims to provide a definition in its language for every term in every language. Consequently, the English Wiktionary contains definitions for many thousands of terms like https://en.wiktionary.org/w/index.php?curid=6100340 for which there's a vanishingly small likelihood of an English-language user of the Wikipedia app (the sole consumer of this endpoint, as far as we know) ever running a search.

TODO
Analyze coverage (for English) for lexemes vs. the current implementation.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptAug 27 2019, 5:49 PM

Change 530259 had a related patch set uploaded (by Mholloway; owner: Michael Holloway):
[mediawiki/services/mobileapps@master] Definitions: Update to use WikibaseLexeme

https://gerrit.wikimedia.org/r/530259

Mholloway updated the task description. (Show Details)Aug 27 2019, 8:55 PM
LGoto triaged this task as Normal priority.Aug 28 2019, 3:35 PM
Mholloway updated the task description. (Show Details)Aug 28 2019, 4:05 PM

Change 530259 abandoned by Mholloway:
Definitions: Update to use WikibaseLexeme

Reason:
Taking this out of the review queue until we're ready to move forward.

https://gerrit.wikimedia.org/r/530259