Projects

Research Initiatives

Photo of a protest regarding the government's response to Covid-19, including posters that say "Quarantine the Sick, Not the Constitution" and "Don't let Covid-19 Infect the Constitution"
Covid-19 Protest
Writing the History of Covid-19: Lessons for the Next Pandemic

Summer 2023 Research Program

Covid-19 Corpus search
Covid-19 Corpus

Our COVID-19 corpus aims aggregates digitized documents related to the the initial phases of the pandemic. They are now divided into individual emails, which can be searched and sorted with the original metadata (from, to, subject, etc.) as well as new metadata we generated using topic modeling and named entity recognition.

Prototypes

A visualization of hundreds of thousands of State Department cables from the 1970s, including still-classified documents.
Spheres of Influence Visualization
Spheres of Influence Visualization

A visualization of hundreds of thousands of State Department cables from the 1970s, including still-classified documents.

Other papers and presentations

Graph of the use of government cables from 1973 to 1979
Detecting and Characterizing Events
Detecting and Characterizing Events

Allison J.B. Chaney, Hannah Wallach, Matthew Connelly, David M. Blei

Experiments from the Lab

Overview of the most redacted figures discussed in the article
America's Most Redacted
America's Most Redacted

Read Our Methodology

Cover photo of the Man vs. Machine page
History in the Making: Man vs. Machine
History in the Making: Man vs. Machine

Read Our Methodology 

Graph showing the "Subject of State Department Communications, from 1974-1979," graphing in order from largest to smallest, Economic/Business, Travel/Exchanges, Political, Social, Military, and Technology
The Subject of State Department Communications, from 1974-1979
Trade and Economic Diplomacy

Read Our Methodology

Matthew Connelly, Nathan Fabius, Raymond Hicks

Graph titled "Percentage of Documents with Redactions by Topic (1961-1968)"
Topic Modeling Official Secrecy
Topic Modeling Official Secrecy

David Allen, Matthew Connelly, Daniel Krasner, Ian Langmore, Thomas Nyberg

Diagram showing the mistakes in categorization made by an AI — false negatives & positives vs. true negatives & positives
Using AI to ID State Secrets
Using AI to ID State Secrets

Renato Rocha Souza, Flavio Codeco Coelho, Matthew Connelly, Rohan Shah