Jump to content

2024:Program/Revision quality algorithms: a subjective view

From Wikimania

Session title: Revision quality algorithms: a subjective view

Session type: Demonstration
Track: Technology
Language: en

Wikimedians can now choose from several machine learning models which provide insight into different aspects of revisions. But which one is right for *your* wiki? We'll take a subjective and user-centric tour of the models and the proposed Automoderator feature.

Description

[edit | edit source]

With the advent of the Automoderator tool, communities will have a tool for automatic reverts at their disposal. But there are more similar algorithms out there, and not all are born equal.

This session will take a subjective and user-centric look at three of the ML models available from WMF's teams, with a case study from the Romanian Wikipedia, in order to familiarize the audience with the pros and cons of the different models.

Session recording: https://www.youtube.com/watch?v=govhrZZjV6s&t=15570

How does your session relate to the event theme, Collaboration of the Open?

The Automoderator tool is developed solely by the WMF, but with Lift Wing, at least in theory, anyone can deploy an ML model for the community to use. By presenting this subject to the community, I hope to foster more collaboration within the community and more openness from the development team.

What is the experience level needed for the audience for your session?

Everyone can participate in this session

Resources

[edit | edit source]

Speakers

[edit | edit source]
  • Strainu
Programmer by formation, Wikipedian for 18+ years.