Analyzing Islamophobic Discourse Using Semi-Coded Terms and LLMs

2025-03-24

AI Summary

This study addresses the global (particularly Western) online propagation of Islamophobia by systematically investigating the usage mechanisms and impacts of semi-encrypted discriminatory terms—such as “muzrat” and “pislam”—on extremist social platforms. Methodologically, it integrates web crawling, large language model (LLM)-based semantic analysis, Google Perspective API toxicity scoring, and BERT-based topic modeling to construct a qualitative lexicon and conduct large-scale empirical analysis. Results demonstrate that LLMs effectively identify out-of-vocabulary (OOV) semi-encrypted slurs; texts containing such terms exhibit significantly higher toxicity than conventional hate speech; and their dissemination is deeply embedded within far-right, conspiracy-theory, and anti-immigration political discourses, coalescing around three core themes: “cultural replacement,” “religious threat,” and “loss of immigration policy control.” The study contributes a deployable detection framework for platform content moderation and provides empirical evidence linking ideologically divergent extremist narratives.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Application Domains: Humanities & Computational Social Science

Application Category

• Web Mining and Content Analysis: Sentiment analysis and opinion mining
• Responsible Web: Measurement, analysis, and circumvention of Web censorship
• Social Networks and Social Media: Media and governance, opinion dynamics, filter bubbles, polarization

Abstract

Islamophobia started evolving into a global phenomenon by attracting followers across the globe, particularly in Western societies. Thus, understanding Islamophobia's global spread and online dissemination is crucial. This paper performs a large-scale analysis of specialized, semi-coded Islamophobic terms such as (muzrat, pislam, mudslime, mohammedan, muzzies) floated on extremist social platforms, i.e., 4Chan, Gab, Telegram, etc. First, we use large language models (LLMs) to show their ability to understand these terms. Second, using Google Perspective API, we also find that Islamophobic text is more toxic compared to other kinds of hate speech. Finally, we use BERT topic modeling approach to extract different topics and Islamophobic discourse on these social platforms. Our findings indicate that LLMs understand these Out-Of-Vocabulary (OOV) slurs; however, measures are still required to control such discourse. Our topic modeling also indicates that Islamophobic text is found across various political, conspiratorial, and far-right movements and is particularly directed against Muslim immigrants. Taken altogether, we performed the first study on Islamophobic semi-coded terms and shed a global light on Islamophobia.

Problem

• Analyzing Islamophobic discourse using semi-coded terms and LLMs
• Assessing toxicity of Islamophobic text compared to other hate speech
• Exploring Islamophobia's link to political and far-right movements

Innovation

• Using LLMs to understand semi-coded Islamophobic terms
• Analyzing toxicity with Google Perspective API
• Extracting topics via BERT topic modeling