The repository contains the code for my master's dissertation titled "Measurement and Role of Emotions in Online Political Communication During the Ukraine War." Essentially, I am developing a novel unsupervised labeling framework that combines Large Language Models (LLMs) and word embeddings into a two-stage labeling process. I validate this method with human coders from Amazon's Mechanical Turk and apply this new tool to analyze President Zelenskyy's Telegram messages during the Ukraine-Russian war.
The dissertation was selected by the Department of Methodology's Program Director as one of the three best examples for future cohorts.