Political “footprints” are vector-based representations of political discourse, where each vector corresponds to a word. These representations are computed using machine learning and pre-trained word embeddings, allowing for a more structurally independent form of political analysis.
This project laid the conceptual and technical groundwork for my paper on post-structuralism in machine learning.
Developed in Python using IBM Watson, GloVe, and TensorFlow.
October 2017
