Modelling the language of extremist communities

Extremist groups develop complex in-group language, also referred to as cryptolects, to exclude or mislead outsiders. Though this is a longstanding and well-documented social norm, it highlights key shortcomings in current natural language processing technologies, e.g. large language models (LLMs), especially when used for content moderation. In this talk, I will describe recent methods, datasets, and models we developed to address these challenges. Our experiments center on two online extremist platforms, Incels and Stormfront, which promote alt-right and misogynistic ideologies, respectively.

Zoom link: https://cam-ac-uk.zoom.us/j/83115049986?pwd=6W5bzFb49HcCbWqz6HR3tRhpVxubTb.1