The Missing Parts: Augmenting Fact Verification with Half-Truth Detection

2025-08-01

AI Summary

Existing fact-checking systems perform binary verification—determining only whether a claim is supported or refuted by evidence—yet fail to detect “half-true” statements: claims that are literally accurate but misleading due to omitted critical context. This work introduces *semi-truth detection* as a novel task and presents *PolitiFact-Hidden*, the first benchmark dataset explicitly designed to evaluate models’ ability to identify missing implicit information. We propose TRACER, a modular framework integrating evidence alignment, implicit intent inference, and causal impact modeling to enable multi-stage reasoning over unstated premises and context dependencies. Experiments demonstrate that TRACER achieves up to a 16-percentage-point improvement in F1-score on the Half-True class over strong baselines, significantly enhancing model sensitivity to deceptive veracity. This advances fact verification from coarse-grained binary judgment toward fine-grained, context-aware credibility assessment.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Fact verification systems typically assess whether a claim is supported by retrieved evidence, assuming that truthfulness depends solely on what is stated. However, many real-world claims are half-truths, factually correct yet misleading due to the omission of critical context. Existing models struggle with such cases, as they are not designed to reason about what is left unsaid. We introduce the task of half-truth detection, and propose PolitiFact-Hidden, a new benchmark with 15k political claims annotated with sentence-level evidence alignment and inferred claim intent. To address this challenge, we present TRACER, a modular re-assessment framework that identifies omission-based misinformation by aligning evidence, inferring implied intent, and estimating the causal impact of hidden content. TRACER can be integrated into existing fact-checking pipelines and consistently improves performance across multiple strong baselines. Notably, it boosts Half-True classification F1 by up to 16 points, highlighting the importance of modeling omissions for trustworthy fact verification.

Problem

• Detecting half-truths in claims missing critical context
• Improving fact verification by modeling omitted information
• Enhancing classification of misleading claims with hidden intent

Innovation

• Introduces half-truth detection task
• Proposes TRACER modular re-assessment framework
• Improves Half-True classification F1 significantly