Have a UROP opening you would like to submit?
Please fill out the form.
Using Natural Language Processing to Analyze Archives of Dictators
17: Political Sciences
F. Daniel Hidalgo
To understand how authoritarian governments work, I am using modern natural language processing techniques and machine learning to analyze the archives of the Brazilian military dictatorship (1964-85). We are using a range of text embedding, entity recognition, and supervised learning methods on millions of documents to characterize the activities of the secret police, opposition figures, and politicians during the period. This rich data will be used to test hypotheses about how authoritarian regimes operate.
Familiarity with R or Python. Experience with natural language processing helpful, but not essential. Not necessary to read Portuguese.