Isaac (WMF)

Isaac Johnson

Senior Research Scientist, Wikimedia Foundation

About me

I joined the the Research team at the Wikimedia Foundation on October 2018 as a Research Scientist. I currently live in the glorious New York City, NY, USA.

My work

My background is in geography and human-computer interaction, with a special focus on understanding (and trying to do something about) how structural inequalities find their way online and into algorithmic systems. Since joining WMF, I have also been heavily involved in research towards better understanding reader needs and behavior, how to model and make predictions about Wikimedia content in a language-agnostic manner, and the impact of external re-use of Wikimedia content.

Disclaimer: I work for or provide services to the Wikimedia Foundation, and this is the account I try to use for edits or statements I make in that role. However, the Foundation does not vet all my activity, so edits, statements, or other contributions made by this account may not reflect the views of the Foundation.

Contact me

• email: [email protected]
• irc: isaacj
• phabricator: @Isaac
• github: @geohci

Tools

A collection of tools that I've built (or helped build) for showcasing some of our research work:

• Language-agnostic content tagging models
• List-building models
• Social media traffic report
• Differential privacy parameter exploration
• Search referral data for Wikipedia
• User scripts for visualizing link data

And specifically, a number of Python packages:

• mwparserfromhtml: parsing Wikipedia HTML (parsoid output)
• mwedittypes: structured analysis of wikitext diffs
• mwtokenizer: word / sentence tokenization for Wikimedia content
• mwsql: parsing Wikimedia SQL dumps

Musings

Various writings about topics relevant to Wikimedia data, research, etc.

• Language modeling:
  • Different ways of categorizing language models at Wikimedia
  • Trade-offs between performance and sustainability in language modeling for Wikimedia
  • Potential content tagging models for Wikimedia
  • Data gaps that inhibit equitable and effective ML for the Wikimedia projects
• Data practices:
  • Various analysis "gotchas" when working with Wikimedia data
  • Aspects to consider when comparing/studying different Wikipedia language editions
  • Standard approaches to various Wikimedia research tasks

Projects

Last updated on 11/16/2023

Active Active projects that I am currently working on: Wikipedia Edit Types NLP Tools for Wikimedia Content Structured data gap List-Building

Completed Completed research projects and reports: Differential privacy for Wikimedia data Knowledge gaps taxonomy Language-agnostic topic classification for Wikipedia articles Report on how information modules on Search impact Wikipedia readership mwsql Python library for processing SQL dumps (Outreachy project) WikiNav tool for exploring clickstream data (Outreachy project) Social media traffic report Wikidata transclusion on English Wikipedia Demographics of Wikipedia readers Survey on the gender of editors Wikidata topic classification (Outreachy project) Evaluating success for content translation (Outreachy project)

Backburner Projects that I've started, but had to put down for the moment: Region inference for Wikipedia articles Sources on Wikipedia Language switching on Wikipedia Sockpuppet detection Entry points External reuse of Wikimedia content Prioritization of Wikipedia articles