Unseen Domain Fake News Detection via Causal Propagation Substructures

ACL ARR 2025 May Submission1281 Authors

Abstract:

The spread of fake news through social media poses significant threats. Recent models using text and graph features have shown promising results in specific fake news detection scenarios. However, these data-driven models heavily rely on training data that share similar distribution with inference data, limiting their applicability to fake news from emerging or previously unseen domains, known as out-of-distribution (OOD) data. Tackling OOD fake news is a challenging yet critical task.

To address this challenge, we propose the Causal Subgraph-oriented Domain Adaptive Fake News Detection (CSoDA) model. CSoDA extracts causal substructures from news propagation graphs that generalize to OOD data, using a graph neural network-based mask generation process. It uses refined training objectives to ensure high-quality subgraphs. It is further powered by contrastive learning for few-shot scenarios, where a limited amount of OOD data is available for training.

Extensive experiments on public social media datasets demonstrate that CSoDA effectively handles OOD fake news detection, achieving a 1.23% to 12.23% accuracy improvement over other state-of-the-art models.

Paper Type: Long

Research Area: Computational Social Science and Cultural Analytics

Research Area Keywords: Computational Social Science and Cultural Analytics, Efficient/Low-Resource Methods for NLP, NLP Applications

Contribution Types: Model analysis & interpretability, NLP engineering experiment, Approaches to low-resource settings

Languages Studied: English, Chinese

Keywords: fake news detection, misinformation, graph neural networks, data mining, causality

Submission Number: 1281