Evolving Hate Speech Online: An Adaptive Framework for Detection and Mitigation

2025-02-15

AI Summary

To address the persistent evolution of hate speech and the detection lag inherent in traditional lexicon-based methods due to static vocabulary lists, this paper proposes an adaptive online detection and mitigation framework. Methodologically, it introduces (1) a word-embedding-driven dynamic lexicon evolution mechanism that enables real-time identification and incremental updating of novel abusive terms and orthographic variants; and (2) a hybrid model integrating BERT-based semantic modeling with dynamic lexicon matching, augmented with context-aware spelling correction and semantic similarity computation. Evaluated on mainstream benchmark datasets, the framework achieves 95% accuracy, significantly improving both detection rate and timeliness for emerging hate speech. It robustly handles challenges including target-group migration and linguistic variation, thereby enhancing adaptability in rapidly shifting sociolinguistic contexts.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Responsible Web: Mitigating online harms, including harassment, spam, and fake reviews
• Web Mining and Content Analysis: Models for Web evolution

Abstract

The proliferation of social media platforms has led to an increase in the spread of hate speech, particularly targeting vulnerable communities. Unfortunately, existing methods for automatically identifying and blocking toxic language rely on pre-constructed lexicons, making them reactive rather than adaptive. As such, these approaches become less effective over time, especially when new communities are targeted with slurs not included in the original datasets. To address this issue, we present an adaptive approach that uses word embeddings to update lexicons and develop a hybrid model that adjusts to emerging slurs and new linguistic patterns. This approach can effectively detect toxic language, including intentional spelling mistakes employed by aggressors to avoid detection. Our hybrid model, which combines BERT with lexicon-based techniques, achieves an accuracy of 95% for most state-of-the-art datasets. Our work has significant implications for creating safer online environments by improving the detection of toxic content and proactively updating the lexicon. Content Warning: This paper contains examples of hate speech that may be triggering.

Problem

• Detect evolving online hate speech
• Adapt to new linguistic patterns
• Improve lexicon-based toxic language detection

Innovation

• Utilizes word embeddings for lexicon updates
• Combines BERT with lexicon-based methods
• Detects intentional misspellings in hate speech