Automatic Fake News Detection: Are Models Learning to Reason?

TL;DR

This paper questions whether fake news detection models truly reason over evidence, revealing that often evidence alone suffices and that including claims can sometimes hinder performance, highlighting issues in current evidence definition.

Contribution

The study critically examines the role of claims versus evidence in fake news detection, challenging assumptions about reasoning and evidence utilization in existing models.

Findings

Evidence alone often yields the highest detection accuracy.

Including claims can be negligible or harmful to model effectiveness.

Current approaches may have issues in defining and utilizing evidence.

Abstract

Most fact checking models for automatic fake news detection are based on reasoning: given a claim with associated evidence, the models aim to estimate the claim veracity based on the supporting or refuting content within the evidence. When these models perform well, it is generally assumed to be due to the models having learned to reason over the evidence with regards to the claim. In this paper, we investigate this assumption of reasoning, by exploring the relationship and importance of both claim and evidence. Surprisingly, we find on political fact checking datasets that most often the highest effectiveness is obtained by utilizing only the evidence, as the impact of including the claim is either negligible or harmful to the effectiveness. This highlights an important problem in what constitutes evidence in existing approaches for automatic fake news detection.