Workshop: Introduction to Latent Semantic Scaling - Nikita Khokhlov (University College Dublin)

14:00-15:30 (GMT) Wednesday, February 4.

Location: SPIRe Boardroom, Newman Building (also live-streamed on Zoom).

Details: This workshop offers an introduction to Latent Semantic Scaling (Watanabe 2021). LSS is a word embeddings method that allows locating texts on the various dimensions, based on their semantic proximity to the pre-defined seed words. We will explore LSS and how researchers use it in measurement, regression and many applications. We will also consider the advantages and drawbacks of LSS, and spend some time learning how to implement it in R.

Prerequisites: Participants need to have installed R, RStudio (or another IDE, such as VSCode or Positron) installed. In addition, participants need to install the following R packages: LSX, quanteda, dplyr, and ggplot2.

About the instructor: Nikita Khokhlov is an Assistant Professor and Ad Astra Fellow in the School of Politics and International Relations, working on comparative authoritarianism, political communication, political behaviour and propaganda with a regional focus on Russia and former Soviet states. His research has been published in the American Journal of Political Science, British Journal of Political Science, Political Communication, Post-Soviet Affairs, and Europe-Asia Studies. He has extensive training in quantitative methods and employs natural language processing, surveys, and experiments in his work.