DeLIVeR: Decomposed Learning for Information-grounded Veracity Recognition via Reinforced Knowledge Graph Exploration
2026-07-20 • Computation and Language
Computation and LanguageArtificial Intelligence
AI summaryⓘ
The authors created a new system called DeLIVeR to help large language models check facts better by breaking down complicated claims into smaller questions. These questions guide the system to find accurate information in a structured knowledge base. They trained the system using a special method that encourages diverse and correct reasoning paths. Tests show DeLIVeR works much better than previous methods on popular fact-checking datasets. This approach also makes it easier to see how the system finds and verifies information.
Large Language ModelsFact-CheckingKnowledge GraphsReinforcement LearningMulti-hop ReasoningPlanner LLMPolicy OptimizationLIAR DatasetFEVER DatasetPolitiFact Dataset
Authors
Cong Hoan Nguyen, Thomas Hoang, Hieu Minh Duong, Long Nguyen
Abstract
Automated fact-checking remains a challenge for Large Language Models (LLMs) due to "query brittleness" in traditional retrieval systems. We propose DeLIVeR (Decomposed Learning for Information-grounded Veracity Recognition), a framework that treats evidence retrieval as a reinforced strategic exploration task. DeLIVeR utilizes a Planner LLM to decompose complex claims into targeted question sets, which are used to traverse structured Knowledge Graphs (KGs) for high-precision evidence. We optimize the Planner's policy using Group Relative Policy Optimization (GRPO) with a reward system prioritizing structural diversity and verdict accuracy. Our evaluation on LIAR, FEVER, and PolitiFact shows that DeLIVeR significantly outperforms state-of-the-art baselines. Using Qwen2.5-7B, our framework achieved peak F1-scores of 83.73, 84.57, and 79.70 respectively, representing a 10-15% improvement over HippoRAG2. By shifting to a reinforced question-planning strategy, DeLIVeR effectively bridges multi-hop reasoning gaps and provides an auditable, transparent path for verifiable misinformation detection.