Description
This paper explores the application of computational visual methods to study the online ecosystems of extremist and violent political actors. Recognizing the challenges posed by vast digital information landscapes, the authors advocate for the integration of image-based computer vision techniques, akin to those used for text analysis. Using an original dataset of images sourced from incel platforms, the study demonstrates the utility of unsupervised deep clustering and supervised object identification. It highlights how these methods can enhance researchers' understanding of extremist imagery, outlining their strengths, limitations, and potential for complementing traditional visual analysis workflows. This innovative approach aims to equip terrorism and extremism analysts with more robust tools to navigate and interpret complex online environments.
Related topic
Online radicalisation, including minors