Contextualized Early Detection of Online Firestorms: A Sequential LLM-Based Approach

2026-07-16

AI Summary

This study addresses the limitations of existing online outrage detection methods, which struggle to capture the dynamic evolution of semantic context within discussion threads, thereby hindering early warning capabilities. To overcome this, the authors propose a dual-mode detection system leveraging large language models (LLMs): a global mode retrospectively classifies entire Reddit threads, while a sequential mode employs a sliding window to continuously estimate key indicators—namely, the proportion of negative sentiment, escalation intensity, and active user count—for real-time early warning. The key innovation lies in the novel use of LLMs as context-aware continuous estimators, transcending the constraints of traditional static features or isolated sentiment scores. Experimental results demonstrate that the approach achieves both high classification accuracy and recall on a balanced Reddit dataset, accurately identifying potentially escalating threads with only a small number of initial comments.

Abstract

Online firestorms are rapid collective escalations of highly negative user-generated content and may cause substantial reputational and economic damage. Existing detectors usually work with volume signals, sentiment scores, or predefined linguistic features. Such signals are useful, but they capture contextual meaning shifts in evolving discussion threads only indirectly. This paper proposes an LLM-based detection system with two operating modes. The first mode classifies complete Reddit threads retrospectively by combining local chunk-level assessments into a thread-level judgment. The second mode processes threads sequentially and issues early warnings when a sliding window exceeds calibrated thresholds. In this mode, the language model estimates three firestorm indicators: negativity share, escalation level, and contributor count. On a balanced Reddit dataset, the global mode achieves strong classification performance, while the early warning mode reaches high recall and detects escalating threads after only a small number of comments and distinct contributors. The results indicate that LLMs can be used not only for static judgment tasks, but also as repeated estimators in context-aware monitoring of social media discourse.

Problem

• online firestorms
• early detection
• contextual meaning
• social media discourse
• dynamic escalation

Innovation

• LLM-based detection
• early warning system
• contextualized monitoring
• online firestorms
• sequential analysis