Detecting Sockpuppetry on Wikipedia Using Meta-Learning

2025-06-12

AI Summary

Detecting malicious sockpuppet accounts on Wikipedia remains challenging in low-resource settings—particularly for novel authors or small-scale sockpuppet groups—due to the inability of existing methods to generalize across diverse, sparse writing styles. To address this, we propose the first meta-learning framework for sockpuppet detection, built upon Model-Agnostic Meta-Learning (MAML). Our approach jointly models textual stylistic features, incorporates multi-task learning, and integrates hand-crafted and embedded metadata features to enable rapid few-shot adaptation to unseen puppet groups. Crucially, we introduce dynamic author-behavior modeling via meta-learning—a novel conceptual and technical contribution to this domain. We also release the first publicly available, fully annotated Wikipedia sockpuppet investigation dataset. Experiments on real-world investigative data demonstrate that our method significantly outperforms strong pretrained baselines across all few-shot settings (1–5 samples per group), achieving superior generalization. This work establishes a new paradigm for trustworthy content governance under data-scarce conditions.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Malicious sockpuppet detection on Wikipedia is critical to preserving access to reliable information on the internet and preventing the spread of disinformation. Prior machine learning approaches rely on stylistic and meta-data features, but do not prioritise adaptability to author-specific behaviours. As a result, they struggle to effectively model the behaviour of specific sockpuppet-groups, especially when text data is limited. To address this, we propose the application of meta-learning, a machine learning technique designed to improve performance in data-scarce settings by training models across multiple tasks. Meta-learning optimises a model for rapid adaptation to the writing style of a new sockpuppet-group. Our results show that meta-learning significantly enhances the precision of predictions compared to pre-trained models, marking an advancement in combating sockpuppetry on open editing platforms. We release a new dataset of sockpuppet investigations to foster future research in both sockpuppetry and meta-learning fields.

Problem

• Detecting malicious sockpuppets on Wikipedia reliably
• Improving adaptability to author-specific behaviors in detection
• Enhancing performance in data-scarce settings using meta-learning

Innovation

• Meta-learning for sockpuppet detection
• Adapts to author-specific writing styles
• Enhances precision in data-scarce settings