Quantifying Indeterminacy in News: A Neutrosophic Method for Assessing Fact-Checking Outcomes

Abstract

Fact-checking news claims often encounters situations beyond simple true/false verdicts, 
involving partial truths, conflicting evidence, or insufficient information. Conventional binary or 
multi-class rating systems in both manual and automated fact-checking struggle to adequately 
represent this nuanced reality, particularly the degree and nature of indeterminacy. This paper 
proposes a novel framework grounded in neutrosophic logic to quantify truth, falsehood, and 
particularly indeterminacy in automated fact-checking outcomes. Neutrosophy explicitly models 
information using independent degrees of Truth (T), Indeterminacy (I), and Falsity (F), offering a 
richer representation than traditional logics. 'Indeterminacy' here encompasses ambiguity, 
vagueness, contradiction, uncertainty, and neutrality. Based on a systematic literature review 
spanning journalism, misinformation studies, and uncertainty modeling, we develop a theoretical 
model and outline a computational procedure for deriving (T, I, F) scores for news claims based on 
aggregated supporting, refuting, and unresolved evidence identified by automated systems. This 
approach allows claims to be assessed not just as true or false, but by how true, how false, and how 
indeterminate they are, given the available evidence, explicitly handling partial truths and 
contradictions. While primarily theoretical at this stage, the proposed framework offers a 
foundation for developing more transparent and nuanced AI fact-checking tools capable of 
communicating the inherent complexities and uncertainties in assessing real-world information, 
bridging neutrosophic theory with journalistic practice

DOI: 10.5281/zenodo.16123010