Maria Antoniak


I’m a Young Investigator at the Allen Institute for AI, where I work in natural language processing and cultural analytics.

I earned my PhD in Information Science from Cornell University, where I was advised by David Mimno. I have a master’s degree in Computational Linguistics from the University of Washington and have worked as a research intern at places like Microsoft Research FATE, Twitter Cortex, Facebook Core Data Science, and Pacific Northwest National Laboratory. I’ve been recognized as a “Rising Star” in both computer science and data science.


Research Interests

📐 Measuring the Reliability of NLP Tools: I’ve shown that popular NLP methods ported to new domains can result in surprising instabilities and biases: for example, word vector similarities require additional stability tests when used to measure social biases.

✨ Modeling Narratives and Values: I’ve used NLP models to study storytelling, framing, and expression of values, investigating questions like where people tell stories online and what values are held by readers of different genres. With collaborators in the humanities, I’ve also studied online reading communities, examining topics like the “classics” according to Goodreads users and the canonization process of books from interwar Paris.

⚕️ Person-Focused NLP for Healthcare: I’ve worked with interdisciplinary teams of clinicians and researchers at Microsoft, Facebook, the Hospital for General Surgery NY, and the Association of American Medical Colleges. My research has focused on modeling the experiences of care-seekers: for example, how postpartum people share and frame their childbirth narratives and how an online community collaboratively makes sense of difficult healthcare decisions. I’ve also built a public facing set of foundations of responsible NLP use for maternal health equity.


News and Upcoming Talks

June 2024 Our work on guiding principles for NLP for maternal healthcare will be presented at FAccT
June 2024 Our work on personalized jargon identification will be presented at NAACL
June 2024 Our work on sensemaking in online discussions of contraception will be presented at ICWSM
May 2024 Invited to give a keynote talk at the Workshop on Reference, Framing, and Perspective at LREC-COLING 2024 in Torino
April 2024 Invited to speak at the Department of Computer Science at the University of Victoria
April 2024 Invited to speak with the NLP Group at the University of British Columbia
Mar 2024 Invited to be a panelist at the IUI Workshop on Human-AI Interaction & Cultural Heritage
Mar 2024 Invited to speak at the Kahlert School of Computing at University of Utah
Mar 2024 Invited to speak at the CS Department at CU Boulder
Feb 2024 Invited to speak at the Max Planck Institute for Software Systems
Feb 2024 Invited to speak at the CS Department at Emory University
Jan 2024 New preprint of our forthcoming study at the Journal of Cultural Analytics and Modernism/modernity
Dec 2023 Invited to give a keynote talk at NLP4DH in Tokyo
Nov 2023 Invited to speak at the Online Seminar in Economics and Data Science hosted by ETH Zurich
Nov 2023 Invited to give a tutorial on "Large Language Models for Humanists: A Hands-On Introduction" with Melanie Walsh at the University of Washington Simpson Center for the Humanities
Nov 2023 Presenting Riveter and our work on values in online book reviews at Text as Data in Amherst
Oct 2023 Invited to speak at the CS department at George Mason University
Oct 2023 Invited to speak at the Quantitative Social Science Colloquium (QSSC) at Princeton University
Oct 2023 Invited to speak at the Machine Learning and Friends Lunch at UMass Amherst


Teaching, Outreach, and Resources

I designed and taught NLP for Cultural Analytics for the Linguistics department at the University of Washington in Winter 2023.

I’m one of the lead organizers for AI for Humanists, a series of tutorials and workshops that guide interdisciplinary researchers in using large language models. In addition to our independent tutorials, I’ve led or co-led sessions at ICWSM, FAccT, Bell Labs, and the popular NLP+CSS 201 tutorial series. I’ve also taught similar public-facing courses for the Hertie School in Berlin, the Brown Institute at Columbia, and the IDEAS Summer School at Northeastern.

I’m the lead builder and maintainer for some cultural analytics tools:


Media


Recent Service


Other Things