Go to ACL ARR 2025 February homepageFactAppeal: Identifying Epistemic Factual Appeals in News Media
Abstract: Detecting how factual claims are supported by appeals to authority, expertise, or evidence is critical for assessing credibility in public discourse. We propose the novel task of Epistemic Appeal Identification, which not only detects whether a statement conveys a factual claim but also reveals how it is anchored by external sources or evidence. To advance research on this task, we present FactAppeal, a manually annotated dataset of 3,226 English-language news sentences capturing both claim factuality and the nuanced epistemic structures underlying these claims. Unlike prior resources that focus solely on claim detection and verification, FactAppeal provides theory-driven, fine-grained annotations of source attribution, quotation method (direct or indirect), and appeal type (e.g., expert testimony, official statements, direct evidence). Our experiments show that generative models consistently outperform encoder-based baselines, underscoring both the complexity of modeling epistemic appeals and the promise of large-scale generative architectures for advancing factuality detection in news media.
Paper Type: Long
Research Area: Discourse and Pragmatics
Research Area Keywords: fact checking, rumor/misinformation detection, communication, argument mining, quantitative analyses of news and/or social media, misinformation detection and analysis
Contribution Types: Data resources
Languages Studied: English
Submission Number: 6875
Loading