University of Maryland

Uncategorized

Announcing ZeitGEIS 2026

Join members of the GEIS research group for our second annual public meeting, where we look back on a year of work on elections and information security and ahead to 2027. Across three panels, GEIS members and invited speakers will share new research on AI and elections, political influence in digital spaces, and socio-technical approaches…

VIOLETS ChatBot in the Field for the 2026 Midterms

During the 2026 midterm elections, GEIS’s VIOLETS (Voter Information from Official Local Election Trusted Sources) chatbot was deployed in the field. Led by Dr. Giovanni Ciampaglia, Dr. Cody Buntain, Dr. Patrick Y. Wu, and PhD student Do Won Kim, VIOLETS answered questions about Maryland elections using only official local election sources. This deployment extended the…

GEIS Members Presenting at POLMETH 2026

Patrick Wu presented new work applying large language models to analyze narratives embedded in tweets, and Masha Krupenkin presented her research on automated scoring of open-ended survey responses and essays, both at POLMETH 2026.

GEIS Team to Present Multiple Projects at IC2S2 2026

Cody Buntain, Giovanni Ciampaglia, and PhD student Do Won Kim will present new research on VIOLETS — GEIS’s AI chatbot for Maryland election information — at IC2S2 2026 in Burlington, VT. The paper builds on the team’s public demonstration of VIOLETS at Maryland Day earlier this year. Masha Krupenkin has been presenting her research on…

New Framework Measures Persuasive Language Across Domains: The Persuasion Index

A new paper co-authored by GEIS’s Julia Mendelsohn — “The Persuasion Index: An Interpretable Framework for Quantifying Rhetorical Strategies Across Domains” (Gong, Wang, Xu, & Mendelsohn) — has been accepted for presentation at the 2026 International Conference on Computational Social Science (IC2S2) in Burlington, VT. The framework offers a way to measure and compare persuasive…

New Preprint: Challenging Partisan Expectations Reduces Political Polarization

Giovanni Ciampaglia and PhD student Do Won Kim have released a new working paper, “Challenging Partisan Expectations Reduces Political Polarization.” The study looks at how correcting people’s mistaken assumptions about the opposing political party can help lower polarization. https://bsky.app/profile/dowonkim.bsky.social/post/3mofy47rpac2b

TikTok Deactivation Study Presented at PaCSS 2026

Cody Buntain and Masha Krupenkin from GEIS, together with Patricia Alegria, Benjamin Guinaudeau, and Joshua A. Tucker, presented new research on TikTok’s recommendation systems at the 2026 Political and Computational Social Science (PaCSS) conference. The presentation is part of the team’s ongoing work examining how platform recommendation algorithms shape the political content users encounter online…

Directions from June 2026

GEIS closed out the semester with new preprints, new courses, and a successful community event. New Preprints on Persuasion and Polarization Do Won Kim and Giovanni Ciampaglia posted a new preprint, “Challenging Partisan Expectations Reduces Political Polarization,” finding that political conversations reduce polarization most when they challenge what people expect about partisan boundaries. Krystal Gong,…

VIOLETS Chatbot Sparks Conversations at Maryland Day 2026

GEIS brought its VIOLETS: Voter Information from Official Local Election Trusted Sources AI chatbot to Maryland Day on April 25, inviting visitors to ask an AI chatbot about Maryland elections. The demo, led by Giovanni Ciampaglia, Cody Buntain, and PhD student Do Won Kim, drew a strong turnout and sparked conversations about how conversational AI…

Directions from May 2026

May kept GEIS in full conference-prep mode while the group also took its research public at Maryland Day. Conference Season Ramps Up Members are heading into a busy stretch of abstract deadlines and presentations. The group submitted several abstracts to IC2S2: Cody Buntain and a team of students and a postdoc have two accepted abstracts,…