Goal: empirically understand gaps in patrolling, specifically the magnitude of the inability for editors to see when Wikipedia articles change because Wikidata items that are transcluded are edited.
Background:
- https://meta.wikimedia.org/wiki/Research:Patrolling_on_Wikipedia/Report#WikiData_and_Wikimedia_Commons
- T151717
General approach:
- Use wbc_entity_usage MariaDB table to determine what items/properties are transcluded in which Wikipedia articles
- For each Wikipedia article, compute the ratio of edits to the article : edits to the Wikidata properties transcluded.
- This is slightly tricky because wbc_entity_usage does not track when the property started to be transcluded. Potentially one could grab the current snapshot of the table and then wait a month and gather the data for that month only.
- Also we would need to understand the potential for new properties to be added to a Wikidata item that would automatically be transcluded -- e.g., via infobox templates.