Page MenuHomePhabricator

Newcomer tasks: SE for kowiki not scored correctly (investigate)
Open, Needs TriagePublic

Description

Several articles on kowiki seem to be assigned Geography.Regions.Africa.Central Africa(100%) incorrectly.

The steps to get suggested edits articles with Geography.Regions.Africa.Central Africa(100%).

  1. On 'kowiki' Special:Homepage select
  2. all filters for Geography topics
  3. copyedit as the difficulty level

The examples below were returned in the result set for the above filters.

Examples:
(1)
https://ko.wikipedia.org/wiki/%EB%A0%88%EA%B3%A0_%ED%85%8C%ED%81%AC%EB%8B%89

Geography.Regions.Africa.Central Africa (100%)
STEM.STEM * (79%)
History and Society.Business and economics (72%)
Culture.Media.Media * (66%)
Culture.Media.Entertainment (33%)
STEM.Engineering (31%)
History and Society.Transportation (22%)

The article in enwiki https://en.wikipedia.org/wiki/Lego_Technic has

STEM.STEM*(67%)
History and Society.Business and economics(58%)
Culture.Media.Media*(31%)
STEM.Engineering(31%)
History and Society.Transportation(26%)
Culture.Media.Entertainment(25%)
STEM.Technology(23%)

(2)
https://ko.wikipedia.org/wiki/%EB%8D%B0%EB%8B%88%EC%8A%A4_%ED%91%B8%EC%8B%A4%EB%A6%B0

Geography.Regions.Africa.Central Africa(100%)
Geography.Regions.Europe.Eastern Europe(97%)
Geography.Regions.Europe.Europe*(95%)
History and Society.History(52%)
Culture.Biography.Biography*(49%)
History and Society.Politics and government(24%)
History and Society.Military and warfare(16%)

And in enwiki https://en.wikipedia.org/wiki/Denis_Pushilin will be scored without Geography.Regions.Africa.Central Africa

Geography.Regions.Europe.Eastern Europe(98%)
Geography.Regions.Europe.Europe*(97%)
Culture.Biography.Biography*(84%)
History and Society.Military and warfare(38%)
Geography.Regions.Asia.Asia*(29%)
History and Society.Politics and government(27%)

(3)
https://ko.wikipedia.org/wiki/%EB%B0%B0%ED%8B%80%EB%A1%9C%EB%B4%87

Geography.Regions.Africa.Central Africa(100%)
STEM.STEM*(66%)
STEM.Technology(63%)
STEM.Engineering(16%)

The scores for enwiki article https://en.wikipedia.org/wiki/Robot_combat

STEM.Technology(95%)
STEM.STEM*(80%)
Culture.Media.Media*(42%)
History and Society.Business and economics(15%)

Also, for all the above articles I looked up the translation (via Google translate) and neither in categories nor in the text Central Africa was not mentioned (as far as I could see it).

Note
The numbers in ORES Thresholds for Geography.Regions.Africa.Central Africa seem to be incomplete and it might be affecting the scoring.

pop ratethresholdprecisionrecall
Geography.Regions.Africa.Central Africa0.00.9< 0.15

Event Timeline

@Etonkovidova -- thank you for detailing this. @revi also noticed this issue, and so did the other ambassadors. Central Africa is the worst-performing topic. I am going to make this task into a subtask of T246909: Follow-up cleanup to topic models, where the researchers are working on improving the models.