Wikimedia Discovery/Meetings/Checkin/2017-09-19

 Topics from the pastEdit

  • Wikimedia Developer Summit 2018 (Jan 22-23): Call for Position Statements, due Friday 9/29
  • WMF All-staff January 25-26
  • There have been questions about our human relevance testing, Trey is writing a blog post to explain the logic behind the tests.

Announcements, Information, QuestionsEdit

  • Quarterly Review for Search Platform (Technology) scheduled for 10/16
  • Erika and Chelsy (and Amanda) went to an interesting product-focused presentation on “Using Machine Learning to Build Great Products” where they used a Problem/Solution/Challenges/Outcome model to present three use cases from LinkedIn, AirBnB and Slack.
  • Consider weekend offsite after all hands in January, may be challenging to get all of “Discovery” together
  • Trey’s blog post about our human search relevance testing went live:
  • Did a few kick-off meetings recently to talk about Structured Data on Commons future work.
  • Deb will drive Discovery check-ins next two weeks, Erika will be out

 Scrum of ScrumsEdit

Are we blocked?

  • None

Are we blocking?

  • None

Other dependencies (in either direction) which don’t need to be called out as “blocked” (e.g. are progressing smoothly, have no urgency, etc.)

  • None

 Discovery NewsEdit

 Quick Quarterly Goals/KPI Update (if needed)Edit

Discovery Roadmap FY 2017/18:

FY 2017-18 Q1 (Jul-Sep) goals:

This status was last updated 2017-09-19. Completed/dropped goals may not be shown.

Tech:Search PlatformEdit

Backend Search:

1. Implement advanced methodologies such as “learning to rank” machine learning techniques and signals to improve search result relevance across language Wikipedias.

  • Perform load tests and A/B tests on new models to make sure they can be safely deployed to production (IN PROGRESS)
  • When ready, deploy newly automated models which match (at a minimum) current performance of manually-configured search result relevance (DONE)

2. Improve support for multiple languages by researching and deploying new language analyzers as they make sense to individual language wikis.

  • Perform research spikes to find new analyzers for different languages (DONE for Japanese and Vietnamese)
  • Test new analyzers to see if they are improvements (Japanese and Vietnamese) (DONE—they are not, alas.)
  • Deploy new / updated analyzers (DONE—none to deploy)
  • Deploy analyzers in progress from last quarter (Hebrew)(WAITING--sooooo close!)

WDQS: Wikidata Query Service goal for this quarter will be working on expanding category search in the query service, while also collecting SPARQL statistics; it will be maintained by Stas and Guillaume to support the continued growth and use of the service; the Analysis team will help with statistics.


Frontend Search:

  • Finish up testing of the Explore Similar feature on search results page
    • Analyze data and gather community feedback (DONE)
    • Request internationalization strings (DONE)
    • Deploy feature (ON HOLD PENDING 2ND TEST)

Portal: Portal team will not have any dedicated goals this quarter other than continuing to maintain the page by fixing critical bugs and performing regular statistics and translations updates.

Maps: The Maps goal for this quarter is to finalize and deploy the new map style, as well as monitoring the service for critical bugs and increasing the frequency of OSM replication.

Analysis: The team will continue to work closely with the Search teams to analyze A/B tests and other assorted data; they will also prototype automated A/B test reports and investigate addition of Continuous Integration to their R codebases.


  • Sep 18 - 29 are all remote weeks when the SF office move happens
  • Oct 9: US Holiday (Labor Day)
  • Nov 11: US Holiday
  • OOO
    • Erika out 9/25-26 for Wikilead
    • Chris out Sept 28-30 for Strange Loop
    • Trey out Sept 28-29, not for Strange Loop
    • Deb out Sep 29, also not for Strange Loop
    • Erika out 10/3-10/4 for Agile Open NorCal conference
    • Erika out 10/13 for Transformative Technology Conference
    • Paul off October 19 - 22 + travel for State of the Map US in Boulder, CO
    • Deb off October 20 for State of the Map US in Boulder, CO
    • Chelsy off October 20 for vacation (instead of Columbus on Oct 9)
    • Chris at Readers Apps off-site Oct 16-20 in Philadelphia
    • Chris at Readers Web off-site Oct 23-27 in Queens, NY
    • Mikhail at Readers Apps off-site Oct 16-20 in Philadelphia
    • Stas at WikidataCon 28-29 Oct in Berlin