Page MenuHomePhabricator

Files with title set get incorrectly added to Category:Files with no machine-readable description
Open, HighPublic

Description

On Commons we use several infobox templates:

(and some less used others)

Both Artwork and Book require either description or title to be set. Now artworks and books that only have the title set (and not description) incorrectly end up in https://commons.wikimedia.org/wiki/Category:Files_with_no_machine-readable_description

You should probably update the code to make use of this title field and not add this category if it's set.

See also https://commons.wikimedia.org/w/index.php?title=Commons:Village_pump&oldid=136612339#Mysterious_categories


Version: unspecified
Severity: normal
URL: https://commons.wikimedia.org/wiki/Category:Files_with_no_machine-readable_description

Details

Reference
bz71950

Event Timeline

bzimport raised the priority of this task from to Needs Triage.Nov 22 2014, 3:47 AM
bzimport added a project: CommonsMetadata.
bzimport set Reference to bz71950.
bzimport added a subscriber: Unknown Object (MLST).

This would mean using fileinfotpl_art_title as a fallback for fileinfotpl_desc. I looked into doing the fallback directly in the template, but because we're using IDs, that can't work.

(In reply to Maarten Dammers from comment #0)

Both Artwork and Book require either description or title to be set. Now
artworks and books that only have the title set (and not description)
incorrectly end up in
https://commons.wikimedia.org/wiki/Category:Files_with_no_machine-
readable_description

I would argue that's correct, at least in theory. A title and a description are different things (and will have different fields in the structured data); having a title is no reason not to add a description as well. For one thing, key concepts should be wikilinked in the description, which practically never happens in the artwork/book titles. Also, the titles are messy - e.g. half the HABS ([[c:Category:Files_from_the_Historic_American_Buildings_Survey]]) images have a "title" which is the concatenation of the actual title and the description; for other files it is a concatenation of the actual title and the source. So there is no reason to exclude these files from a description cleanup drive, IMO. For the same reasons, I don't think it makes sense for the template to pretend its title field to be a description.

In practice, we might want to say such images are less problematic than ones with neither title nor description and the category should be limited to those, at least initially; I'll defer to you and Guillaume on that.

Gilles subscribed.

Currently CommonsMetadata does not make any distinction between the title field of the template and the EXIF title; and the EXIF title tends to be crap.