Rochana Chaturvedi

My research spans different aspects of natural language processing, graph-structured modeling, and causal machine learning. My research goal is to develop methods that extract structure and meaning from complex, unstructured text to address problems of real societal consequence. I focus on two broad themes:

• Extracting structure from text to support decisions. I develop methods to model temporally and causally grounded knowledge from unstructured, longitudinal text to improve decision support and social systems; for example, extracting patient timelines to predict early disease risk, or analyzing language to understand how it reflects and shapes human behavior, such as group affiliation and polarization.
• Making language model behavior trustworthy. I study how design and alignment choices in LLMs give rise to bias and vulnerabilities in high-stakes decision-making, and build NLP systems that are not only accurate but also fair, efficient, reliable and secure.

I am motivated by compelling problems, and some of the domains that I have studied include clinical NLP, computational social science, climate-risk and critical infrastructure decision support, and behavioral finance.

I am currently a Postdoctoral Fellow at the Kellogg School of Management at Northwestern University and the Northwestern Institute on Complex Systems (NICO). I previously served as a Postdoctoral Fellow in the Mathematics and Computer Science Division at Argonne National Laboratory. I earned my Ph.D. and Master’s in Computer Science from the University of Illinois Chicago.

Before my doctoral studies, I served as an Assistant Professor of Computer Science at Keshav Mahavidyalaya, University of Delhi, and worked as an Associate Software Engineer at Objective Systems Integrators. I hold a prior Master’s degree in Computer Applications from Guru Gobind Singh Indraprastha University, and a Bachelor’s degree in Physics from the University of Delhi.