Grants
- NSF HNDS-R: Understanding Drivers of Trust in Cryptocurrency Using Big Data and Ethnographic Approaches (September 2023 – August 2026)
- NSF SAI-P: Overcoming Barriers to User-Centered Infrastructure Planning with System Modeling and Natural Language Processing (September 2022 – August 2024)
- Received 2021-2022 PSU President’s Diversity Mini-Grants (April 2022)
Publications
Featured Research
Community Insights with LLMs
Infrastructure
Large Language Models can be used to solicit and analyze public comments on infrastructure projects and to prepare participatory modeling work. This helps infrastructure planners understand how different communities experience and judge project impacts. We work on practical approaches for planners and explore the reliability, context-richness, and causal reasoning of LLMs.
Cryptocurrency
We combine big data analysis of social media, using Large Language Models, with ethnographic methods (netnography, participant observation, surveys and interviews= to understand phenomena that are almost exclusively observable online, such as trust in cryptocurrency during catastrophic market downturns. Our work leads to new insights into trust and new research methods for social sciences.
Fairness in AI
Dialect Diversity in Text Summaries
Social media posts provide a compelling, yet
challenging source of data of diverse perspectives from many socially salient groups. Automatic text summarization algorithms make
this data accessible at scale by compressing
large collections of documents into short summaries that preserve salient information from
the source texts. This work improves the quality of summaries generated from social media data in terms of their ability to represent not only salient but also diverse perspectives.
PDF – Data
Completed Projects
Fuzzy Operational Cognition of Safety Culture (FOCOS)
FOCOS is a self-guided online training solution that is specifically designed for the offshore energy industry. Its mission is to prevent accidents and to make offshore operations more resilient by training users on best practices of high-reliability safety culture.
Demo Website