About the Project

Sensing the increasing polarization in America, in 2020 I embarked on a project aimed to quantify and visualize common ground amongst the 100 senators. The idea was to not only understand what was on each senator's mind but how it related with other senators. The project mines the recent tweets of the senators and uses various NLP techniques to identify topics and commonality across the senators. Each week I identify the three most pressing topics discussed by the senators using K-means clustering. The clusters are created using word vectors representing the recent tweets. The senators that belong to each clusterer are then presented on a map describing the state they represent. In addition I also score each senator on how similar they are to both others in their own party and opposing party. This is done by computing the distance between the word vectors. Finally, I represent the most discussed topics across both parties in a word cloud graphic.