Traverse the category graph when searching for media on Commons (Community Wishlist/W400)
Long-term opportunity
Description
When I do a mediasearch on Commons various fields get queried (see here), including category ... so if I do a search for "dogs" one of the reasons an image might be returned is it's in https://commons.wikimedia.org/wiki/Category:Dogs
What search does not do, however, is look in subcategories. Ideally if I search for "dogs" then images in https://commons.wikimedia.org/wiki/Category:Collies would also be candidates for inclusion in the search results, but right now they are not (unless they match in some other way)
This would involve:
- including parent category names in the search index, up to some pre-defined depth
- making sure that changes to categories/subcategories (and not just File pages themselves) get propagated into the search index
- re-calibrating mediasearch to include search signals from parent-category matches
Note that the WMF tried this before with "depicts", and found that so much noise got added to the search index that results got worse rather than better
Assigned focus area
Unassigned
Type of wish
System change
Tags
Affected users
People searching on Commons
Other details
- Created: 11:52, 24 June 2025
- Last updated: 16:23, 10 October 2025
- Author: CParle (WMF) (talk)
Voting