Status Update (May 2, 2018)


  • We've started work on JADE in earnest, and the prototype is deployed to the beta cluster where it's available for testing and tool development.
  • Draft topic prerequisites are mostly falling into place, so we should be able to get the initial model deployed this month.
  • New, dynamic ORES support table shows up-to-date information about our progress for each wiki: https://tools.wmflabs.org/ores-support-checklist/
  • ORES is served from its own cluster, which gave us a tremendous benefit in both performance and stability.
  • More ORES support for Arabic, Bengali, Catalan, Hungarian, Latvian, Swedish Wikipedia


T121719: [Epic] Write paper about ORES as a socio-technical probe
T188123: Present about draft topic model at Wikimedia Research Showcase.
T188124: Build slide deck about AI at Wikimedia for Policy People
T190464: Discuss surfacing ORES for AFC/NPP

Draft topic

T123327: Train/test draft topic model (new article routing AI)
T185147: Host Google-News-word2vec.bin publicly
T185896: OneVsRest Classification for revscoring
T188445: Implement word2vec featurevector in revscoring
T189364: Investigate word2vec memory issues with multiprocessing

Article quality

T127470: Deploy item quality classification model for Wikidata
T158437: Change ORES rules to send all events to new "/precache" endpoint
T174684: Article quality campaign for Persian Wikipedia
T181522: Fix response processing logic in drafttopic.fetch_page_wikiprojects
T181820: Rename "wikiclass" to "articlequality"
T187927: Drop "non bot" condition from ORES changeprop rules

Edit quality

T130279: Deploy ORES filters to Spanish Wikipedia
T131669: Complete edit quality campaign for Arabic Wikipedia
T145394: Deploy ORES filters in es.wikibooks
T163005: Complete Latvian Wikipedia editquality campaign
T163006: Train/test damaging & goodfaith models for Latvian Wikipedia
T166235: Flagged revs approve model to fiwiki
T167968: Complete edit quality campaign for Hungarian Wikipedia
T174560: Enable ORES filters for svwiki
T174878: Edit quality campaign for Bengali Wikipedia
T182012: Deploy ORES filters to Simple Wikipedia
T185903: Train/test damaging and goodfaith model for Hungarian Wikipedia
T186711: Edit quality campaign for Bengali Wikisource
T187503: Bengali Wikisource not recognized as a wikimedia wiki
T187732: Train/test damaging/goodfaith model for Catalan Wikipedia
T187742: I broke merge_labels utility so bad
T187771: Edit quality campaign for Catalan Wikipedia
T187836: [Epic] Audit of pending ORES GUI deployments
T188270: Simplify hewiki, cswiki, plwiki, svwiki
T188535: JSON error in building models
T188781: Simplify azwiki's labeling campaign
T189710: Train and test damaging/goodfaith model for arwiki
T192362: Nulls appear in labeled data (merge_labels issue)

Judgment and dialogue engine

T153152: Design JADE data storage schema
T176333: Deploy JADE prototype in Beta Cluster
T182055: Vagrant role for JADE
T187216: Build prototype JADE extension
T189938: Create mediawiki.org article for Extension:JADE


T168073: Switch ORES to dedicated cluster
T168455: [Epic] Implement code generation for model makefile maintenance
T169246: Stress/capacity test new ores* cluster
T171851: Reimage ores* hosts with Debian Stretch
T180587: Scream properly when threshold stats fetch is failing
T181071: Cache ORES virtualenv within versioned source
T181159: Migrate ORES extension threshold config from old to new syntax
T182799: Make sure ORES is compatible with stretch
T184072: Rebuild ORES models on Stretch
T184073: Provision a Stretch box we can use to build ORES models
T184074: Verify that all enchant/spelling dictionaries are available on Stretch. Port if needed.
T184077: ORES MediaWiki-Vagrant roles should be ported to Stretch
T184135: Rebuild ORES wheels on Stretch
T184451: Selenium tests for ORES
T184938: Special:RecentChanges broken on Jenkins slaves
T185901: Preliminary deployment of ORES to new cluster
T186453: Fix or decide on edge cases of Makefile
T186557: WikiLabels OAuth handshake doesn't work with HTTPS
T187914: New precache endpoint isn't reporting its metrics correctly
T188777: makefile generate makes impossible rules when no model is available
T189605: Beta cluster ORES is emitting statsd errors
T189781: Reboot oresrdb
T189948: Clean up old backward compatibility settings of $wgOresModels
T189954: Build ORES support checklist
T191321: Remove deprecated hosts from ORES scap config
T191435: Highlight wikis that has their campaign almost completed
T191808: Fix the ORES-extension grafana dashboard
T192038: Upgrade ORES mwparserfromhell dependency to 0.5.1


T179874: Specter of revscoring 1 haunting ORES
T181163: Revscoring tune does not recognize a set of labels as target
T188755: Update ORES requirements to support revscoring 2.2.0
T192917: Rebuild all models for revscoring 2.2.2

ORES extension

T181334: Split Cache.php to different services
T181335: Redesign onRecentChange_save hook handler for ORES
T181892: Rewrite Stats.php
T183199: ORES Extension refactoring
T183266: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith]
T185534: Clean up ScoreLookup implementations

Written by awight on May 2 2018, 8:05 PM.
"Like" token, awarded by MarcoAurelio.

Event Timeline