Saturday 11:00–11:45 in Hörsaal 3

ML and populism

Limor Gultchin

Audience level:


They say AI and Machine Learning do and will transform every aspect of our lives. I tried to give it a shot by using the tools we all know to understand better the rise of populism across the US and Europe, possibly the most acute political problem of the last few years, and the years to come. Here's what I learned.


The literature and political science research on populism was for many years broken down across regions and stuck in arguments among scholars themselves about how to define the term populism itself. I compiled a data set of world populist, based on scholarly mentions, tracked some usual paths to their progress, analysed manifestos and compared to agreeable non-populist baseline to start and define what a populist speech is. Using word2vec, t-sne, PCA, TFITF, etc. I started to get a better sense of what populist speech might be, and aspire to build a scoring/predicting system for populist speech, to help the field, and the public debate, get a better understanding of what it is we're talking about when talking about populism.

Subscribe to Receive PyData Updates