From Words to Actions: A Comprehensive Approach to Identifying Incel Behavior on Reddit
ISEF · 2025 Behavioral and Social Sciences
Overview
The incel (involuntary celibate) community is a radicalized online subculture. In this study, I focus on the popular social media platform Reddit and analyze the activity patterns and comments of 14,000 users to characterize the mechanics of online radicalization and offer tools for identifying at-risk individuals before they become deeply entrenched in incel ideologies. My work utilizes systematic data collection to construct a 7000 user incel-positive group and an equal-sized incel-negative group who share similar activity levels in the neutral subreddits frequented by the positive users. I devise a hybrid method that combines both language analysis on users’ comments and their behavior patterns across different subreddits. I find that utilizing language analysis alone yields a classification accuracy of 76%, while using only the subreddit activity patterns of users achieves an accuracy of 79%. Remarkably, my hybrid method blends both approaches and reduces false positives and negatives by cross-referencing data to achieve 99.8% accuracy, improving 20% over the state-of-the-art classification solutions. This work is important because it can spot individuals at risk of joining harmful groups early on, allowing for timely support. While focused on Reddit and incels specifically, this hybrid method can be adapted across other online communities and platforms, and the insights from this work can guide platform moderators and policymakers in creating safer online spaces.
Competition history
- ISEF 2025
Resources
Related projects
ISEF · 2020
Using Natural Language Processing and Linguistic Insights to Combat Cyberbullying and Toxicity in Social Media
ISEF · 2022
SuiSensor: A Novel, Low-Cost Machine Learning System for Real-Time Suicide Risk Identification and Treatment Optimization via Computational Linguistics
ISEF · 2026
Digital Expressions of Discontent: A Computational Analysis of Anti-Work Reddit Communities and Evolving Work Narratives
ISEF · 2024
Detecting and Quantifying Emotional Loneliness Risk in the Digital Age: An Innovative Lexicon-Based Text Analysis Method Combining Deep Learning and Rule-Based Classification
Closest projects by meaning, across every fair and year in the corpus.
Source: Regeneron International Science and Engineering Fair