Comparing the old and new digital far-right: the “hate focus” on Stormfront and the Daily Stormer

• In Event: Exploring Recent Trends in Crime and Punishment

Abstract

I introduce a new method for analyzing large document collections: Inductive-Deductive Topic Modeling. I use IDTM to compare the "hate focus" on the formerly and currently most popular far-right sites, Stormfront and the Daily Stormer.

IDTM has four steps. 1. Qualitative induction: drawing on preliminary research and/or reviews of the literature, I draft word dictionaries that represent topics of interest (for example, in research on the far right, a key topic is gender). 2. Computational induction: with one or more topic dictionaries drafted (Gender = women, men), I construct word embeddings based on the document collection, to find related keywords (soyboy, incel). 3. Qualitative refinement: to validate topics, I construct and review concordances -- text snippets with keywords in their semantic context. 4. Quantitative deduction: I now count finalized dictionary words within the full dataset (has talk about gender increased over time?).

Author

• Isak Ladegaard, University of Illinois Urbana-Champaign