DARE: Dialectical Agentic Reasoning for Structured Knowledge Fact Checking

Yifei Li; Xiaohan Zheng; Wentao Qian; Liansheng Zhuang

cs.CLSep 12, 2026

DARE: Dialectical Agentic Reasoning for Structured Knowledge Fact Checking

Authors: Yifei Li, Xiaohan Zheng, Wentao Qian, Liansheng Zhuang

Organizations: University of Science and Technology of China

Abstract

Structured knowledge fact checking aims to determine the truthfulness of natural language claims by reasoning over structured evidence. Recent program-generation approaches leverage large language models (LLMs) to generate executable graph reasoning programs, achieving strong performance on structured knowledge fact checking benchmarks. However, these methods remain limited by invalid relation generation, single-path reasoning that lacks self-correction, and biased evidence assessment that tends to overestimate supporting signals. We propose Dialectical Agentic Reasoning (DARE), a multi-agent framework that formulates structured knowledge fact checking as an iterative retrieve-reason-reflect process. DARE integrates relation-grounded evidence retrieval to constrain reasoning to valid structures, dialectical bidirectional verification to evaluate evidence from both supporting and refuting perspectives, and confidence-driven meta-reflection to dynamically determine whether additional evidence exploration is necessary. Extensive experiments demonstrate the effectiveness of DARE in structured knowledge fact checking, with an 8B backbone achieving 88.12% accuracy and matching or surpassing GPT-4o-based program-generation baselines, which attests to the efficacy of dialectical agentic reasoning in eliciting the latent reasoning capabilities of LLMs.

Figures & tables

Figure 1: Interaction among the retriever, reasoner, and controller agents.

Figure 2: Overview of the DARE framework for structured knowledge fact checking. The retriever iteratively extracts relation-grounded evidence from the knowledge graph. The reasoner performs dialectical bidirectional reasoning over supporting and refuting perspectives to produce evidence-aware assessments. The controller applies confidence-driven meta-reflection to determine whether to terminate with a final prediction or continue iterative reasoning with expanded entities.

Without Evidence
Method	Evidence	Strategy	1-Hop	Existence	Multi-Hop	Conjunction	Negation	Acc.
BERT	w/o	full	69.64	61.84	70.06	63.31	63.62	65.20
BlueBERT	w/o	full	60.03	59.89	57.79	60.15	58.90	59.93
Flan-T5	w/o	0-shot	62.17	55.29	60.67	69.66	55.02	62.70
ChatGPT	w/o	12-shot	—	—	—	—	—	68.48
With Evidence

Table 1: Accuracy (%) on the FactKG dataset. Methods are categorized by evidence usage and training strategy. The best results are in bold . DARE uses Qwen3-8B as the backbone model.

Variant	Acc.	$\Delta$
DARE (Full)	88.12	—
w/o Reasoner’s Dialectical Reasoning	63.20	$-$ 24.92
w/o Controller’s Meta-Reflection	70.11	$-$ 18.01
w/o Retriever’s Semantic Filtering	72.21	$-$ 15.91

Table 2: Ablation study on FactKG. Each variant removes the core mechanism of one agent.

\Delta

shows the accuracy change relative to the full model.

Figure 3: Impact of the maximum iteration limits (

T

) on the overall accuracy.

Figure 4: Impact of the confidence gap threshold (

\tau

) on the overall accuracy.

Claim: Aaron Boogaard was from Canada where one of the languages spoken is French.

Label: True Type: Conjunction / Multi-claim

Baseline: PGR (GPT-4o)

MATCH(Aaron_Boogaard, placeOfBirth, ?x) AND MATCH(?x, language, French_language) .

Error: The strict programmatic syntax fails to map "from" to the exact valid KG paths in a single pass, leading to an execution failure.

Prediction: False

\times

DARE (Ours)

Table 3: Case study on a multi-hop claim.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 5: Graph patterns for conjunction and multi-hop claims in FactKG.

Type	Written	Colloquial-Model	Colloquial-Presup	Total
One-hop	2,106	15,934	1,580	19,620
Conjunction	20,587	15,908	602	37,097
Existence	280	4,060	4,832	9,172
Multi-hop	10,239	16,420	603	27,262
Negation	1,340	12,466	1,807	15,613
Total	34,552	64,788	9,424	108,764

Appendix

Table 4: FactKG Dataset Statistics

Method	Evidence	Strategy	Quant-Free	FOL-Quant	Entity-Comp.	Count-Quant	Count-Comp.	Acc.
GPT-4o	w/o	0-shot	73.25	59.50	71.77	60.69	56.15	62.47
DeepSeek-V3.2	w/o	0-shot	66.50	52.47	64.72	56.05	52.46	56.70
BERT	w/o	full-training	78.50	66.22	48.70	81.08	63.18	68.26
BERT	w/	full-training	92.25	66.37	52.80	80.43	61.42	70.12
GEAR(KG)	w/	full-training	91.17	68.45	55.12	80.81	66.32	72.50
KG-GPT ∗	w/	few-shot	78.00	47.89	44.22	48.85	48.96	51.75

Appendix

Table 5: Accuracy (%) on the Fact-FOLX-KG dataset. A superscript

*

denotes a DeepSeek-V3.2-based baseline variant; unmarked LLM-based baseline variants use GPT-4o. For our method, we report results obtained with Qwen3-8B. The best results among the included methods are in bold .

Automated Fact-Checking

Automated fact-checking detects checkable claims, retrieves evidence, and assesses whether that evidence supports or refutes them.

See all 139 papers

DeLIVeR: Decomposed Learning for Information-grounded Veracity Recognition via Reinforced Knowledge Graph Exploration
Jul 20, 2026Cong Hoan Nguyen, Thomas Hoang, Hieu Minh Duong +1LLM Grounding RL for Language Model Reasoning
University of Louisville, Louisville KY 40292, USA. · Denison University, Granville, Ohio 43023, USA.
Multi-Sourced, Multi-Agent Evidence Retrieval for Fact-Checking
Feb 27, 2026Shuzhi Gong, Richard O. Sinnott, Jianzhong Qi +3Automated Fact-Checking Evidence Retrieval
The University of Melbourne, Melbourne, VIC, Australia · Data61, CSIRO, Sydney, NSW, Australia · MBZUAI, Abu Dhabi, UAE
DeepFact: Co-Evolving Benchmarks and Agents for Deep Research Factuality
Mar 6, 2026Yukun Huang, Leonardo F. R. Ribeiro, Momchil Hardalov +3Automated Fact-Checking AI Agent Benchmarks
Duke University · Amazon AGI · Boston University

Jul 20, 2026J/K move · Enter open · S save

DeLIVeR: Decomposed Learning for Information-grounded Veracity Recognition via Reinforced Knowledge Graph Exploration

Cong Hoan Nguyen, Thomas Hoang, Hieu Minh Duong, Long Nguyen

University of Louisville, Louisville KY 40292, USA. · Denison University, Granville, Ohio 43023, USA.

LLM Grounding RL for Language Model Reasoning Automated Fact-Checking

Open paper AskPDF arXiv

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.

Abstract

Automated Fact-Checking

Explore similar work

DeLIVeR: Decomposed Learning for Information-grounded Veracity Recognition via Reinforced Knowledge Graph Exploration

Multi-Sourced, Multi-Agent Evidence Retrieval for Fact-Checking

DeepFact: Co-Evolving Benchmarks and Agents for Deep Research Factuality