[01:05:24] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES, 10ORES: [Discuss] Process for enabling ORES beta feature on big wikis - https://phabricator.wikimedia.org/T138250#2440400 (10Johan) Then I don't think it should be that controversial at this stage, as long as we explicitly remind everyone... [01:12:04] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES, 10ORES: [Discuss] Process for enabling ORES beta feature on big wikis - https://phabricator.wikimedia.org/T138250#2394743 (10Legoktm) At the same time we do this, I think we should be starting social outreach to enable RC patrolling on the... [01:15:18] PROBLEM - ORES web node labs ores-web-03 on ores.wmflabs.org is CRITICAL: CRITICAL - Socket timeout after 10 seconds [01:17:37] RECOVERY - ORES web node labs ores-web-03 on ores.wmflabs.org is OK: HTTP OK: HTTP/1.1 200 OK - 338 bytes in 8.574 second response time [01:20:51] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES, 10ORES: [Discuss] Process for enabling ORES beta feature on big wikis - https://phabricator.wikimedia.org/T138250#2440445 (10Ladsgroup) Also this is the announcement I send every time I enable it in a wiki: [[ https://meta.wikimedia.org/wi... [01:55:48] 10Revision-Scoring-As-A-Service-Backlog, 10ORES, 07Wikimedia-log-errors: Model contains an error: ValueError: Failed to process datasource.wikibase.revision.parent.item_doc: Expecting value - https://phabricator.wikimedia.org/T139660#2440496 (10Ladsgroup) [02:08:09] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES: Build an entry point to store scores in ORES extension - https://phabricator.wikimedia.org/T131785#2178505 (10Legoktm) What is the use case for this? [03:19:44] 06Revision-Scoring-As-A-Service, 10MediaWiki-extensions-ORES, 10Wikimedia-Site-requests, 13Patch-For-Review: Deploy ORES review tool in ptwiki - https://phabricator.wikimedia.org/T139692#2440551 (10Dereckson) 05Open>03Resolved [06:06:32] 06Revision-Scoring-As-A-Service, 10ORES: Don't load models into memory on web workers - https://phabricator.wikimedia.org/T139407#2431228 (10Ladsgroup) It is becoming more worrying: https://ganglia.wikimedia.org/latest/graph.php?r=month&z=xlarge&h=scb1001.eqiad.wmnet&m=cpu_report&s=by+name&mc=2&g=mem_report&c=... [06:06:39] 06Revision-Scoring-As-A-Service, 10ORES: Don't load models into memory on web workers - https://phabricator.wikimedia.org/T139407#2440636 (10Ladsgroup) p:05Triage>03High [07:42:58] (03PS1) 10Ladsgroup: Highlight the whole row in EnhancedChangesList [extensions/ORES] - 10https://gerrit.wikimedia.org/r/297956 [08:49:05] (03PS2) 10Ladsgroup: Highlight the whole row in EnhancedChangesList [extensions/ORES] - 10https://gerrit.wikimedia.org/r/297956 [11:05:43] (03CR) 10Thiemo Mättig (WMDE): [C: 031] "Code looks good. Not tested myself." [extensions/ORES] - 10https://gerrit.wikimedia.org/r/297956 (owner: 10Ladsgroup) [12:44:34] 06Revision-Scoring-As-A-Service, 10revscoring, 10rsaas-editquality: Train/test reverted model for enwiktionary - https://phabricator.wikimedia.org/T138630#2441279 (10jberkel) Thanks! What are the next steps to actually tag new enwikt edits with this model? [15:12:11] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES: Improve "Hide good edits from recent changes" label - https://phabricator.wikimedia.org/T139754#2441630 (10He7d3r) [18:05:16] 06Revision-Scoring-As-A-Service, 10revscoring, 10rsaas-editquality: Train/test reverted model for enwiktionary - https://phabricator.wikimedia.org/T138630#2442408 (10Halfak) I'll see if I can get it deployed today :) [19:13:10] https://gerrit.wikimedia.org/r/#/c/298039/ [19:16:33] * halfak pushing cswiki and enwiktionary to staging [19:54:36] 06Revision-Scoring-As-A-Service, 10ORES, 10rsaas-editquality: Deploy cswiki and enwiktionary reverted models - https://phabricator.wikimedia.org/T139789#2442725 (10Halfak) [19:54:58] 06Revision-Scoring-As-A-Service, 10ORES, 10rsaas-editquality: Deploy cswiki and enwiktionary reverted models - https://phabricator.wikimedia.org/T139789#2442739 (10Halfak) https://ores.wmflabs.org/v2/scores/cswiki/reverted/2345678 works https://ores.wmflabs.org/v2/scores/enwiktionary/reverted/2345678 works... [19:56:18] 06Revision-Scoring-As-A-Service, 10revscoring, 10rsaas-editquality: Train/test reverted model for enwiktionary - https://phabricator.wikimedia.org/T138630#2442743 (10Halfak) Deployed now! See https://ores.wmflabs.org/v2/scores/enwiktionary/reverted/2345678 for an example scoring of https://en.wiktionary.org... [20:28:34] halfak o [20:28:44] Hi GhassanMas! [20:29:09] how are you doing [20:30:03] just want to give feedback regarding the arwiki campaign [20:32:36] Doing good. [20:32:43] Cool. What's up? [20:32:53] me as well (: [20:33:48] regarding the campaign what I found that the typical reason for bad edits are: [20:34:29] 1) the editor write a his/her signature after an edit [20:35:07] 2) writing in English chara instead of Arabic [20:35:54] 3) removing the editor stuff like "[[ ]]" [20:35:54] 10[2] 10https://meta.wikimedia.org/wiki/ [20:36:32] those are good faith I suppose [20:37:01] Seems like it would be, yeah. [20:37:09] Is the english usually relevant to the article? [20:37:37] what do you mean by that [20:40:16] GhassanMas, I could imagine english language bad-faith vandalism on arwiki articles. [20:40:30] OR it could be good edits -- just in the wrong language. [20:41:52] yeah I go for your second propose " I think it's just the editor doesn't know that he is currently typing English" it's common when you have two or language for keyboard [20:42:12] 06Revision-Scoring-As-A-Service, 10revscoring, 10rsaas-editquality: Train/test reverted model for enwiktionary - https://phabricator.wikimedia.org/T138630#2442931 (10jberkel) Great, thanks! I'll have a play with it. [20:42:38] bad faith typically happens in conversational articles like sport teams [20:42:41] fans stuff [20:42:53] Gotcha. Curse words and racial slurs? [20:43:03] Maybe sexual stuff? [20:43:59] maybe i catch 2-3 sexual stuff it's rarely common so far [20:44:33] Gotcha. WHat would you say the frequency is? 2-3 out of 2000 or more common than that? [20:45:00] so far yes [20:46:43] Hmm... Hopefully we can get a few more observations of that. [20:47:36] I will be more specific for the rest of the samples [20:48:48] the last thing is that I am concerned about is when changing facts like " how many time does won the trophy " [20:49:35] "how many times the team won the trophy" [20:51:47] I imagine those will be hard to catch, but do your best to label them honestly and we'll do our best to model it. [20:53:46] yeah I imagine that, however some facts like the birth day of a typical figure would it be easier catch ? [20:54:27] could we use wikidata [20:55:24] but how do we make sure it's correct data there in wikidata in the first place :D [20:55:48] GhassanMas, yeah. Definitely will be hard. But on the other hand, you can look at the predictions ORES will make as "this needs review" [20:56:00] Many edits that are perfectly OK ought to be reviewed regardless. [20:56:09] And changing a birthday is a good thing to flag. [20:56:21] If vandalism happens that way often, the model should pick it up. [21:01:29] Gotcha [21:47:33] \o/ got the issues worked out with the plwiki resampling stuff. [22:33:27] 10Revision-Scoring-As-A-Service-Backlog, 10ORES, 15User-Johan: Announce deployment of reverted model to svwiki community - https://phabricator.wikimedia.org/T138624#2443156 (10Johan) 05Open>03Resolved a:03Johan [[ https://sv.wikipedia.org/wiki/Wikipedia:Bybrunnen#ORES:_Hj.C3.A4lp_beh.C3.B6vs | It was d... [23:34:51] 10Revision-Scoring-As-A-Service-Backlog, 10MediaWiki-extensions-ORES: Improve "Hide good edits from recent changes" label - https://phabricator.wikimedia.org/T139754#2441630 (10Legoktm) Yeah...how about "probably"?