Exploring Thematic Coherence in Fake News

TL;DR

This paper investigates the use of topic models to analyze thematic coherence in fake news, revealing that fake news exhibits greater deviation in coherence compared to real news across multiple datasets.

Contribution

It introduces a novel approach using topic models to quantify thematic coherence differences between fake and real news across various domains.

Findings

Fake news shows higher thematic deviation in coherence.

Topic models effectively differentiate fake from real news.

Results validated on seven diverse datasets.

Abstract

The spread of fake news remains a serious global issue; understanding and curtailing it is paramount. One way of differentiating between deceptive and truthful stories is by analyzing their coherence. This study explores the use of topic models to analyze the coherence of cross-domain news shared online. Experimental results on seven cross-domain datasets demonstrate that fake news shows a greater thematic deviation between its opening sentences and its remainder.