Page MenuHomePhabricator
Create Task
Maniphest T202573

articlequality scores a very short article really high in Persian Wikipedia
Open, LowPublic
Actions

Description

This pretty short article is scored 4.39 (FA) in ORES. This result seems wrong.

Related Objects

Event Timeline

Ladsgroup created this task.Aug 22 2018, 7:45 PM
Restricted Application added a project: artificial-intelligence. · View Herald TranscriptAug 22 2018, 7:45 PM
Restricted Application added a subscriber: Aklapper. · View Herald Transcript
• Mardetanha subscribed.Aug 22 2018, 7:54 PM
Alfa80 subscribed.Aug 22 2018, 8:03 PM
awight moved this task from Unsorted to Maintenance/cleanup on the Machine-Learning-Team board.Aug 29 2018, 4:24 PM
awight renamed this task from articeqaulity scores a very short article really high in Persian Wikipedia to articlequality scores a very short article really high in Persian Wikipedia.Sep 26 2018, 6:27 PM
Halfak subscribed.Oct 3 2018, 2:36 PM

I bet this has something to do with our training set including Redirects or other types of short wikitext articles rated highly. We should look through our training set to see if we have many example of highly-rated short articles.

Halfak added a comment.Oct 3 2018, 2:52 PM

Here are our FA quality observations. It looks like the first two are excessively short. Could these be mis-labeled?

Qualitytext lengthrev_id
FA269822209198
FA278322931379
FA926821595797
FA926821595797
FA1792722790480
FA1857219544614
FA1925521422786
FA2031322332851
FA2059322236347
FA2170922271457
FA2275421265916
FA2275522925983
FA2358722343509
FA2358922762032
FA2375022960348
FA2376322357296
FA2426719416403
FA2566121352991
FA2597622267990
FA2651221276368
FA2791821298974
... snip ...
FA14280322345176
FA14293822951851
FA14562622914865
FA14653722846950
FA14654022329970
FA14865622287634
FA15367222122809
FA15367222122809
FA15408322848012
FA15413622284243
FA16009721935665
FA16010422304823
FA16082322908335
FA16083422156021
FA16186522286873
FA16898822083811
FA16898822083811
FA19799722289502
FA20701822278332
Ladsgroup added a comment.Oct 8 2018, 3:46 PM

It's quite likely but because of two mislabels, we should not miss-classify future cases that off.

Ladsgroup triaged this task as Low priority.Nov 26 2018, 5:44 PM
Ladsgroup merged a task: T212055: ORES predicts a surprisingly high score for a stub article on fawiki.Dec 16 2018, 4:23 PM
Ladsgroup unsubscribed.Apr 17 2019, 7:14 PM
• ACraze moved this task from Maintenance/cleanup to Backlog/ORES on the Machine-Learning-Team board.Jan 19 2021, 10:19 PM
Content licensed under Creative Commons Attribution-ShareAlike (CC BY-SA) 4.0 unless otherwise noted; code licensed under GNU General Public License (GPL) 2.0 or later and other open source licenses. By using this site, you agree to the Terms of Use, Privacy Policy, and Code of Conduct. · Wikimedia Foundation · Privacy Policy · Code of Conduct · Terms of Use · Disclaimer · CC-BY-SA · GPL · Credits