Page MenuHomePhabricator

Make ORES dependency solving upstreamable
Closed, DeclinedPublic

Description

Our dependency solving and extraction framework seems to be exemplary, and would be a valuable contribution to upstream work. This could take a few forms, the current thoughts are:

  • Express extraction as a scikit-learn transformer and pipeline step.
  • Keep all of our customizability wrt. context, feature injection, cache, config.
  • Decouple from scoring.model
  • Decouple from rev_id -> MediaWiki, the input and extractor are generalized.
  • Include an example that does something simple with e.g. OSM.

Draft,
https://github.com/wiki-ai/revscoring/compare/sklearn_plugin?expand=1

Event Timeline

Unlicking this interesting cookie in case anyone else wants to take a shot.

Ladsgroup raised the priority of this task from Low to Needs Triage.
Ladsgroup moved this task from Unsorted to New development on the Machine-Learning-Team board.
elukey subscribed.

The ML team is focusing on https://wikitech.wikimedia.org/wiki/Machine_Learning/LiftWing, the replacement of ORES. Please re-open if you feel that something is missing, but we'd prefer to add any new functionality only on Lift Wing.