Laure Thompson

As a research software engineer, Laure Thompson builds tools and creates methodologies that enable scholars to use machine learning, natural language processing, and statistical methods for studying humanities collections at scale. She collaborates with Princeton scholars on longer-term CDH-sponsored projects. She also advises and consults with the Princeton community on a wide range of computational and data-intensive projects.

Before coming to Princeton, Laure was an assistant professor in the College of Computer and Information Sciences at the University of Massachusetts Amherst. She received her PhD in Computer Science from Cornell University in 2020 where she was advised by David Mimno. During her time at Cornell, she also completed a graduate minor in classical archaeology advised by Caitie Barrett.

Laure’s expertise is in natural language processing with particular interests in cultural analytics, model interpretability, and data-centric interventions. Her research focuses on understanding what computational models actually learn and how we can intentionally change what they learn. She works with a wide range of cultural heritage corpora: from texts of science fiction novels and medieval manuscripts to images of avant-garde journals and magical gems from the ancient Mediterranean.

She has presented and published work in range of venues that reflects the interdisciplinary nature of her research. Including:

• with David Mimno, "Humanities and Human-Centered Machine Learning" in Human-centered Machine Learning (forthcoming).
• with David Mimno, "What Do Contextualized Representations Represent?" Text as Data (TADA) 2021. poster, preprint
• with David Mimno, "Finding Speculative Fiction in HathiTrust" ACH 2021. slides
• with David Mimno, "Authorless Topic Models: Biasing Models Away from Known Structure" COLING 2018. Won Best Paper Award: Best NLP Engineering Experiment. paper, slides, code
• with David Mimno, "Computational Cut-Ups: The Influence of Dada" in the Journal of Modern Periodical Studies Vol. 8, No. 2 (2017). preprint