Mining User-aware Multi-relations for Fake News Detection in Large Scale Online Social Networks

TL;DR

This paper introduces Us-DeFake, a novel fake news detection model that leverages a dual-layer graph of users and news to incorporate user credibility signals, significantly improving detection accuracy in large-scale social networks.

Contribution

The paper proposes a dual-layer graph model and a new detection method, Us-DeFake, which effectively fuses user credibility signals into news features for improved fake news detection.

Findings

Us-DeFake outperforms all baseline models on real-world datasets.

Incorporating user credibility signals enhances detection accuracy.

The model scales efficiently to large social networks.

Abstract

Users' involvement in creating and propagating news is a vital aspect of fake news detection in online social networks. Intuitively, credible users are more likely to share trustworthy news, while untrusted users have a higher probability of spreading untrustworthy news. In this paper, we construct a dual-layer graph (i.e., the news layer and the user layer) to extract multiple relations of news and users in social networks to derive rich information for detecting fake news. Based on the dual-layer graph, we propose a fake news detection model named Us-DeFake. It learns the propagation features of news in the news layer and the interaction features of users in the user layer. Through the inter-layer in the graph, Us-DeFake fuses the user signals that contain credibility information into the news features, to provide distinctive user-aware embeddings of news for fake news detection. The training process conducts on multiple dual-layer subgraphs obtained by a graph sampler to scale Us-DeFake in large scale social networks. Extensive experiments on real-world datasets illustrate the superiority of Us-DeFake which outperforms all baselines, and the users' credibility signals learned by interaction relation can notably improve the performance of our model.