Email: ndeas AT cs DOT columbia DOT edu
Office: Schapiro CEPSR 7LW1
[ CV ][ Github ][ Google Scholar ] [ LinkedIn ]
Research Interests
Linguistic Biases in LLMs
Much of the text found online used for pre-training language models reflects White Mainstream English, leading to models that are less capable of understanding other dialects and language varieties of English such as African American Language. This research direction seeks to evaluate and mitigate linguistic biases to help create models capable of understanding a wider diversity of language.
Modeling Social Psychology
As language models continue to be applied in psychological research and mental health settings, they must be capable of interpreting linguistic expressions of emotion or mental health. This research direction draws on theories and findings in social psychology and psycholinguistics to embue models with better understanding of human interpersonal behavior.
Political Perspectives and Polarization
With increased interaction on social media and a diverse array of media sources, it is important to ensure that language models fairly learn from and aid productive public discourse. This research direction aims to evaluate the political alignment of language models, ensure fair representation of perspectives on issues, and build models capable of understanding nuances of political speech and rhetoric.
Publications
[NLP: NLP and Computational Linguistics , P: Psychology, PS: Political Science]
[NLP,PS] Summarization of Opinionated Political Documents with Varied Perpsectives
Nicholas Deas, Kathleen McKeown (Preprint)[NLP,P] MASIVE: Open-Ended Affective State Identification in English and Spanish
Nicholas Deas, Elsbeth Turcan, Iván Pérez Mejía, Kathleen McKeown (EMNLP 2024)[NLP] PhonATe: Impact of Type-written Phonological Features of African American Language on Generative Language Modeling Tasks
Nicholas Deas, Jessi Grieser, Xinmeng Hou, Shana Kleiner, Tajh Martin, Sreya Nandanampati, Desmond U Patton, Kathleen McKeown (COLM 2024)[NLP] How Negativity and Policy Content Drive the Spread of Political Messages
Jeffrey A. Fine, D. Hudson Smith, Cierra Oliveira, Nicholas Deas, Spencer Shellnutt, Riley Stotzky, Rachel Clyburn (Journal of Information Technology & Politics)[NLP,PS] Evaluation of African American Language Bias in Natural Language Generation
Nicholas Deas, Jessi Grieser, Shana Kleiner, Desmond Patton, Elsbeth Turcan, Kathleen McKeown (EMNLP 2023)[NLP,P] I just want to matter: Examining the role of anti-mattering in online suicide support communities using natural language processing
Nicholas Deas, Robin Kowalski, Sophie Finnell, Emily Radovic, Hailey Carroll, Chelsea Robbins, Andrew Cook, Kenzie Hurley, Natalie Cote, Kelly Evans, Isabella Lorenzo, Kelly Kiser, Gabriela Mochizuki, Meredith Mock, and Lyndsey Brewer (Computers in Human Behavior 2023)[P] Protection Motivation Theory and intentions to receive the COVID-19 vaccine.
Robin M. Kowalski, Nicholas Deas, Noah Britt, Emily Richardson, Sophie Finnell, Kelly Evans, Hailey Carroll, Andrew Cook, Emily Radovic, Tanner Huyck, Isabella Parise, Chelsea Robbins, Hannah Chitty, and Sophie Catanzaro. (AIMS Public Health 2022)[PS] Partisan Differences in Politicians’ Rhetoric about COVID-19, and Why These Messages Spread Online.
Nicholas Deas, Cierra Oliveira, Riley Stotzky, and Leah Terry (MPSA 2022)[NLP, PS] Using Natural Language Processing to Automate Detection of Targeted Attacks in Political Tweets.
Nicholas Deas, Jacob Sargent, and Spencer Shellnut (MPSA 2019)[NLP, PS] How Characteristics of Members of the House of Representatives and the Political Environment Affect the Use of Political Attacks on Twitter.
Jacob Sargent, Spencer Shellnut, and Nicholas Deas (MPSA 2019)MPSA Best Undergraduate Research Award