Page MenuHomePhabricator

[L] Exclude tables and lists from section alignment-based image suggestions
Closed, ResolvedPublic

Description

Section-level image suggestions (SLIS) leverages Section-Topics plus section alignment.
The exclusion of table and lists is currently implemented in the former component and the same behavior shall apply to the latter.

Iteration 1

Iteration 2

  • refactor duplicate logic into a shared component
NOTE: depending on time constraints, we may leave iteration 2 behind. This slightly increases the technical debt, since duplicate code is less maintainable and elegant.
NOTE: we’ll consider direct refactoring rather than duplication if duplication turns out to be non-trivial.

Report

Snapshot: 2023-03-06

T330848: [XL] Exclude sections with non-standard tables and lists resulted in the following counts of pages that have suggestions:

wikibeforeafterdifference
arwiki97,47187,26810,203
bnwiki30,86928,8382,031
cswiki145,642133,15312,489
eswiki386,910319,36967,541
idwiki83,93975,1868,753
ptwiki192,969164,36128,608
ruwiki342,254306,71335,541
overall1,280,0541,114,888165,166 (- 13%)

T323505: [L] Exclude sections-tables from having section topics resulted in the following counts of pages that have suggestions:

wikibeforeafterdifference
arwiki87,26845,34241,926
bnwiki28,83816,14512,693
cswiki133,15371,62261,531
eswiki319,369178,181141,188
idwiki75,18630,67044,516
ptwiki164,36191,36372,998
ruwiki306,713175,109131,604
overall1,114,888608,432506,456 (- 45%)

Summary

The total decrease of pages with at least one suggestion is - 52% .

Details

ReferenceSource BranchDest BranchAuthorTitle
repos/structured-data/section-image-recs!4standard-filtermainmfossatiPort the standard wikitext list/table filter from section topics
repos/structured-data/section-image-recs!3table-filtermainmfossatiPort the HTML table filter from section topics
Customize query in GitLab

Event Timeline

CBogen renamed this task from Exclude tables and lists from section alignment-based image suggestions to [L] Exclude tables and lists from section alignment-based image suggestions .Mar 8 2023, 5:49 PM

Merge request for the HTML table filter at https://gitlab.wikimedia.org/repos/structured-data/section-image-recs/-/merge_requests/3, ping @matthiasmullie for review.
I'll start working on the standard tables and lists filter straight away.

Iteration 1 is all merged!
Kept a note for iteration 2, as we currently don't have capacity. Closing.