From Code To Counteraction: The Promise And Challenges Of Extremism Detection Algorithms

• In Event: Poster Session 1

Abstract

With the proliferation of extremist propaganda in digital spaces, online platforms rely on content detection algorithms to mitigate the risks of radicalization and violence. Advancements in machine learning techniques, such as natural language processing and deep learning, have led to researchers' developing new tools to identify and remove extremist content. Through a systematic review of existing literature (n = 76), this study assesses the capabilities and limitations of these algorithms in reducing user engagement with extremist propaganda. Particular attention will be paid to the datasets, classification techniques, and validation methods deployed by researchers in virtual environments, thus providing insights into the efficacy of machine learning-based approaches in countering extremist propaganda. Additionally, the study examines the impact of contextual factors, such as linguistic nuances and cultural differences, on algorithmic performance, highlighting the importance of adapting detection models to diverse online environments. By exploring novel approaches in algorithm design and training, the study aspires to contribute to the development of more robust and adaptive content detection systems, thereby fostering safer digital environments.

Author

• Kilhah St Fort, CUNY Lehman College