[Paper Review] Fake News Detection with Deep Diffusive Network Model.

Jiawei Zhang, Limeng Cui+2 more|arXiv (Cornell University)|May 22, 2018

Misinformation and Its Impacts61 references76 citations

TL;DR

This paper proposes FakeDetector, a deep diffusive network model that jointly learns representations of news articles, creators, and subjects using explicit and latent textual features to improve fake news detection. Evaluated on a real-world dataset, it outperforms state-of-the-art models, demonstrating superior performance in identifying deceptive content through joint modeling of multi-entity relationships.

ABSTRACT

In recent years, due to the booming development of online social networks, fake news for various commercial and political purposes has been appearing in large numbers and widespread in the online world. With deceptive words, online social network users can get infected by these online fake news easily, which has brought about tremendous effects on the offline society already. An important goal in improving the trustworthiness of information in online social networks is to identify the fake news timely. This paper aims at investigating the principles, methodologies and algorithms for detecting fake news articles, creators and subjects from online social networks and evaluating the corresponding performance. This paper addresses the challenges introduced by the unknown characteristics of fake news and diverse connections among news articles, creators and subjects. Based on a detailed data analysis, this paper introduces a novel automatic fake news credibility inference model, namely FakeDetector. Based on a set of explicit and latent features extracted from the textual information, FakeDetector builds a deep diffusive network model to learn the representations of news articles, creators and subjects simultaneously. Extensive experiments have been done on a real-world fake news dataset to compare FakeDetector with several state-of-the-art models, and the experimental results have demonstrated the effectiveness of the proposed model.

Motivation & Objective

• To address the challenge of detecting fake news in online social networks, where deceptive content spreads rapidly and affects societal trust.
• To model the complex, diverse connections among news articles, creators, and subjects, which are often overlooked in existing detection methods.
• To develop an automatic credibility inference model that captures both explicit and latent features from textual content for improved detection accuracy.
• To evaluate the model's performance against state-of-the-art approaches on a real-world fake news dataset to validate its effectiveness.

Proposed method

• FakeDetector constructs a deep diffusive network model that simultaneously learns representations of news articles, creators, and subjects using textual features.
• The model extracts both explicit features (e.g., linguistic patterns) and latent features (e.g., embeddings from deep neural networks) from news content.
• It leverages the structural relationships among entities—articles, creators, and subjects—by modeling their interactions through a diffusive message-passing mechanism.
• The deep architecture enables hierarchical feature learning across multiple layers, capturing complex patterns in fake news propagation.
• The model is trained end-to-end to optimize credibility inference, with loss functions designed to distinguish real from fake news.
• A detailed data analysis informs the design of feature extraction and the architecture of the diffusive network to handle unknown characteristics of fake news.

Experimental results

Research questions

• RQ1 How can joint modeling of news articles, creators, and subjects improve fake news detection performance compared to isolated entity analysis?
• RQ2 What combination of explicit and latent textual features best captures the deceptive nature of fake news?
• RQ3 To what extent does the deep diffusive network architecture enhance representation learning for credibility inference?
• RQ4 How does FakeDetector compare to state-of-the-art models in detecting fake news on real-world social media data?

Key findings

• FakeDetector achieves superior performance in fake news detection compared to several state-of-the-art models on a real-world dataset.
• The joint modeling of articles, creators, and subjects significantly improves detection accuracy by capturing inter-entity relationships.
• The integration of both explicit and latent textual features enhances the model’s ability to detect deceptive content.
• The deep diffusive network architecture effectively learns hierarchical representations that are robust to the diverse and evolving characteristics of fake news.
• Extensive experiments confirm that the proposed model is effective in identifying fake news with high reliability.
• The model demonstrates strong generalization on unseen data, indicating its potential for real-world deployment in social media monitoring.