Page MenuHomePhabricator

Pull and normalize data from enwiki AIV
Closed, ResolvedPublic

Description

Per T415103: WE 4.8.3 Patrolling metrics from enwiki SPI/AIV:

From initial feedback from English Wikipedia patrollers, it seems that Administrator Intervention against Vandalism (AIV) and Sockpuppet Investigations (SPI) are promising starting points, due to their semi-structured nature and their nature as reporting venues for rule-breaking.

This ticket covers the technical work needed to pull and structure this data. It'll probably:

  • Scrape and store all revisions from AIV, reducing the number of needed API calls and allowing for offline processing
  • Apply a heuristic to classify all revisions. This should be transparent and also be saved if possible as all classifications within a definition should never change
  • Manage the ability to also classify all exceptions to the heuristic and save these decisions
  • Perform some kind of data visualization/aggregation and/or export some kind of csv based on classifications (tbd)

Subject to change as the technical problems are worked through.

Acceptance Criteria:

  • This should be re-runnable, as if the data is valuable we would want to re-run for post-intervention metrics
  • As these pages have no major privileged access requirements, this tool/visualizer should be runnable without additional privileges and ideally not require complex engineering support