Unseen Fake News Detection Through Casual Debiasing

2025-03-06

AI Summary

Existing fake news detection methods suffer from poor cross-domain generalization due to domain bias in training data. Method: This paper proposes FNDCD, a causal-driven debiasing framework—the first to incorporate causal analysis into fake news detection. FNDCD mitigates domain-specific confounding effects via confidence-based reweighting of classification losses and enforces propagation-structure regularization on graph neural networks (GNNs) to suppress spurious correlations between events and domains. The architecture integrates multi-layer perceptrons with GNNs for joint feature learning and graph-based reasoning. Contribution/Results: Evaluated on multiple real-world datasets with non-overlapping news domains, FNDCD achieves an average 8.7% improvement in F1-score over state-of-the-art methods. It significantly enhances robustness against unseen events and cross-domain fake news, demonstrating superior generalization capability under domain shift.

Technology Category

• Application Domains: Misinformation & Fake News
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The widespread dissemination of fake news on social media poses significant risks, necessitating timely and accurate detection. However, existing methods struggle with unseen news due to their reliance on training data from past events and domains, leaving the challenge of detecting novel fake news largely unresolved. To address this, we identify biases in training data tied to specific domains and propose a debiasing solution FNDCD. Originating from causal analysis, FNDCD employs a reweighting strategy based on classification confidence and propagation structure regularization to reduce the influence of domain-specific biases, enhancing the detection of unseen fake news. Experiments on real-world datasets with non-overlapping news domains demonstrate FNDCD's effectiveness in improving generalization across domains.

Problem

• Detecting unseen fake news on social media
• Addressing domain-specific biases in training data
• Enhancing generalization across non-overlapping news domains

Innovation

• Causal debiasing for unseen fake news detection
• Reweighting strategy based on classification confidence
• Propagation structure regularization for domain generalization