arxiv: 2604.18362 · v1 · submitted 2026-04-20 · 💻 cs.CL · cs.IR

Recognition: unknown

ArbGraph: Conflict-Aware Evidence Arbitration for Reliable Long-Form Retrieval-Augmented Generation

Bohui Fang, Qingying Niu, Ruiyang Ren, Wayne Xin Zhao, Yuhao Wang

Authors on Pith no claims yet

Pith reviewed 2026-05-10 04:48 UTC · model grok-4.3

classification 💻 cs.CL cs.IR

keywords retrieval-augmented generationconflict resolutionevidence graphlong-form generationfactual consistencyhallucination reductionatomic claims

0 comments

The pith

ArbGraph improves factual consistency in long-form RAG by resolving conflicts among atomic claims in a graph before generation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper introduces ArbGraph to handle the problem of noisy and contradictory evidence in retrieval-augmented generation for long-form outputs. The approach breaks down documents into atomic claims, builds a graph showing support and contradiction links, and uses iterative arbitration to weigh credibility before the model generates text. This separation of concerns aims to give the generator a cleaner set of facts. Readers would care because it targets a key failure mode in RAG systems where conflicting sources lead to hallucinations or incomplete answers.

Core claim

ArbGraph decomposes retrieved documents into atomic claims and organizes them into a conflict-aware evidence graph with explicit support and contradiction relations. It then applies an intensity-driven iterative arbitration mechanism that propagates credibility signals through evidence interactions to suppress unreliable and inconsistent claims before final generation, providing a coherent evidence foundation for downstream long-form RAG.

What carries the argument

The conflict-aware evidence graph with support and contradiction relations, processed via an intensity-driven iterative arbitration mechanism that propagates credibility signals to suppress inconsistent claims.

If this is right

ArbGraph improves factual recall and information density on LongFact and RAGChecker benchmarks across multiple LLM backbones.
It reduces hallucinations and sensitivity to retrieval noise.
Performance gains are most evident under conflicting or ambiguous evidence conditions.
Evidence validation is separated from text generation to supply a coherent foundation for long-form outputs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same atomic-claim graph approach could be tested on multi-document summarization to check whether conflict resolution improves consistency beyond RAG.
Adding source metadata such as dates or reliability scores to the graph edges might address temporal or authority-based inconsistencies the current model does not cover.
If the arbitration scales efficiently, it could serve as a standard preprocessing filter in production RAG systems handling large evidence sets.

Load-bearing premise

Decomposing documents into atomic claims and modeling their support and contradiction relations in a graph resolves factual conflicts without significant information loss or introduction of new errors.

What would settle it

A controlled test on benchmarks with deliberately injected contradictions where ArbGraph shows no gain or a drop in factual accuracy compared to standard RAG would falsify the central claim.

Figures

Figures reproduced from arXiv: 2604.18362 by Bohui Fang, Qingying Niu, Ruiyang Ren, Wayne Xin Zhao, Yuhao Wang.

**Figure 2.** Figure 2: Sensitivity analysis of key parameters. Moderate values of 𝜏𝑞, 𝑀, and 𝑘 achieve a balance between evidence coverage, structural connectivity, and arbitration stability. 4.4 Analysis of Graph Construction Factors We conduct the following sensitivity analyses on the LongFact benchmark, where claim-level annotations enable fine-grained evaluation of how graph construction affects factual coverage and evidence… view at source ↗

**Figure 3.** Figure 3: Comparison of Scheduling Policies. ArbGraph’s conflict-aware scheduling improves factuality and evidence density. 4.4.6 Effectiveness of Conflict-Aware Scheduling. We analyze the impact of different scheduling strategies on the arbitration process. While heuristic-based scheduling policies such as HardFirst and Easy-First rely on static ordering rules, ArbGraph employs a dynamic, conflict-aware scheduling… view at source ↗

**Figure 4.** Figure 4: Comparison of standard RAG vs. ArbGraph. Standard RAG fails due to noise and unresolved conflicts, while ArbGraph resolves conflicts via evidence-level arbitration. 4.5.2 Case Study. We analyze a challenging query: “How is the United States related to the East Asia Summit (EAS)?” ( [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗

read the original abstract

Retrieval-augmented generation (RAG) remains unreliable in long-form settings, where retrieved evidence is noisy or contradictory, making it difficult for RAG pipelines to maintain factual consistency. Existing approaches focus on retrieval expansion or verification during generation, leaving conflict resolution entangled with generation. To address this limitation, we propose ArbGraph, a framework for pre-generation evidence arbitration in long-form RAG that explicitly resolves factual conflicts. ArbGraph decomposes retrieved documents into atomic claims and organizes them into a conflict-aware evidence graph with explicit support and contradiction relations. On top of this graph, we introduce an intensity-driven iterative arbitration mechanism that propagates credibility signals through evidence interactions, enabling the system to suppress unreliable and inconsistent claims before final generation. In this way, ArbGraph separates evidence validation from text generation and provides a coherent evidence foundation for downstream long-form generation. We evaluate ArbGraph on two widely used long-form RAG benchmarks, LongFact and RAGChecker, using multiple large language model backbones. Experimental results show that ArbGraph consistently improves factual recall and information density while reducing hallucinations and sensitivity to retrieval noise. Additional analyses show that these gains are evident under conflicting or ambiguous evidence, highlighting the effectiveness of evidence-level conflict resolution for improving the reliability of long-form RAG. The implementation is publicly available at https://github.com/1212Judy/ArbGraph.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

ArbGraph adds a pre-generation graph of atomic claims with support/contradiction edges and intensity-driven propagation to clean evidence, but the gains on LongFact and RAGChecker could easily come from basic filtering rather than true conflict resolution.

read the letter

The paper's core move is to pull evidence validation out of the generator entirely. It breaks retrieved documents into atomic claims, wires them into a graph with explicit support and contradiction links, then runs an iterative credibility propagation that down-weights inconsistent claims before any text is produced. That separation is cleaner than most current RAG pipelines, which either stuff more context in or try to fix contradictions while generating. The experiments report better factual recall, higher information density, and fewer hallucinations across several backbones on the two standard long-form benchmarks, and the code is public, which helps.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces ArbGraph, a pre-generation framework for long-form RAG that decomposes retrieved documents into atomic claims, organizes them into a conflict-aware evidence graph with explicit support and contradiction relations, and applies an intensity-driven iterative arbitration mechanism to propagate credibility signals and suppress unreliable claims before generation. Experiments on LongFact and RAGChecker benchmarks with multiple LLM backbones report consistent gains in factual recall and information density alongside reductions in hallucinations and retrieval-noise sensitivity, with additional analysis under conflicting evidence.

Significance. If the central claims hold after addressing verification gaps, ArbGraph would represent a useful contribution to reliable long-form RAG by explicitly separating evidence-level conflict resolution from generation. The public code release supports reproducibility, and the focus on pre-generation arbitration addresses a practical pain point in noisy retrieval settings.

major comments (2)

[§3 and §4] §3 (ArbGraph Framework) and §4 (Experiments): The central claim that the evidence graph plus iterative arbitration produces a cleaner claim set than raw retrieval hinges on accurate atomic claim decomposition and support/contradiction edge labeling, yet no quantitative audit (e.g., precision/recall of decomposition fidelity or edge-label accuracy on the evaluation sets) is provided. Without this, observed gains on LongFact/RAGChecker could be explained by simple filtering rather than conflict resolution.
[§4.3] §4.3 (Analysis under conflicting evidence): The intensity-driven propagation is asserted to suppress noise without amplifying decomposition or labeling errors, but no error-propagation analysis or ablation isolating the arbitration step from baseline filtering is reported. This leaves the load-bearing assumption that the mechanism reliably corrects rather than compounds upstream errors unverified.

minor comments (2)

[Figure 1 and §3.2] Figure 1 and §3.2: The graph construction notation (nodes as claims, edges as support/contradiction) would benefit from a concrete worked example showing how a multi-fact sentence is split and labeled.
[§4.1] §4.1: Clarify the exact prompting templates used for claim decomposition and relation labeling, as these are implementation-critical for reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for your detailed review and constructive feedback on our manuscript. We appreciate the opportunity to clarify and strengthen our work on ArbGraph. Below, we provide point-by-point responses to the major comments.

read point-by-point responses

Referee: [§3 and §4] §3 (ArbGraph Framework) and §4 (Experiments): The central claim that the evidence graph plus iterative arbitration produces a cleaner claim set than raw retrieval hinges on accurate atomic claim decomposition and support/contradiction edge labeling, yet no quantitative audit (e.g., precision/recall of decomposition fidelity or edge-label accuracy on the evaluation sets) is provided. Without this, observed gains on LongFact/RAGChecker could be explained by simple filtering rather than conflict resolution.

Authors: We agree that a quantitative audit of atomic claim decomposition fidelity and support/contradiction edge labeling accuracy would strengthen the validation of the framework components. The current manuscript emphasizes end-to-end gains in factual recall, information density, and hallucination reduction on LongFact and RAGChecker. To address the concern that gains may stem from simple filtering, we will add in revision an ablation comparing full ArbGraph against a baseline using claim decomposition followed by non-graph credibility filtering. We will also include precision/recall metrics on a sampled subset of claims via human or LLM-based judgment to quantify decomposition and labeling quality. revision: yes
Referee: [§4.3] §4.3 (Analysis under conflicting evidence): The intensity-driven propagation is asserted to suppress noise without amplifying decomposition or labeling errors, but no error-propagation analysis or ablation isolating the arbitration step from baseline filtering is reported. This leaves the load-bearing assumption that the mechanism reliably corrects rather than compounds upstream errors unverified.

Authors: We acknowledge that an explicit error-propagation analysis and ablation isolating iterative arbitration from baseline filtering would better verify the mechanism. Section 4.3 shows gains under conflicting evidence, but lacks a dedicated comparison. In revision, we will add an ablation with a single-pass intensity threshold variant (no iterative propagation) to demonstrate the iterative step's contribution. We will also include a sensitivity analysis discussing how the intensity-driven thresholds mitigate potential upstream errors rather than amplify them. revision: yes

Circularity Check

0 steps flagged

No circularity: ArbGraph is an empirical algorithmic framework

full rationale

The paper proposes a new pre-generation arbitration pipeline (claim decomposition into atomic units, construction of a support/contradiction graph, and intensity-driven iterative propagation) and validates it via experiments on LongFact and RAGChecker. No equations, parameter-fitting steps, or predictions appear in the provided text; the central claims rest on benchmark results rather than reducing to inputs by definition or self-citation. The method is self-contained and externally falsifiable.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The central claim relies on the effectiveness of these new structures and mechanisms, which are introduced without prior independent validation in the abstract.

axioms (1)

domain assumption Documents can be decomposed into independent atomic claims without loss of critical context.
This is invoked in the initial decomposition step described in the abstract.

invented entities (2)

Conflict-aware evidence graph no independent evidence
purpose: Organizes atomic claims with explicit support and contradiction relations for arbitration.
New data structure proposed in the framework.
Intensity-driven iterative arbitration mechanism no independent evidence
purpose: Propagates credibility signals through evidence interactions to suppress unreliable claims.
Core algorithmic contribution for conflict resolution.

pith-pipeline@v0.9.0 · 5552 in / 1394 out tokens · 66893 ms · 2026-05-10T04:48:06.258476+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

51 extracted references · 21 canonical work pages · 8 internal anchors

[1]

Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi. 2023. Self-rag: Learning to retrieve, generate, and critique through self-reflection. InThe Twelfth International Conference on Learning Representations

2023
[2]

Maciej Besta, Nils Blach, Ales Kubicek, Robert Gerstenberger, Michal Podstawski, Lukas Gianinazzi, Joanna Gajda, Tomasz Lehmann, Hubert Niewiadomski, Piotr Nyczyk, et al. 2024. Graph of thoughts: Solving elaborate problems with large language models. InProceedings of the AAAI conference on artificial intelligence, Vol. 38. 17682–17690

2024
[3]

Shehzaad Dhuliawala, Mojtaba Komeili, Jing Xu, Roberta Raileanu, Xian Li, Asli Celikyilmaz, and Jason Weston. 2024. Chain-of-verification reduces hallucination in large language models. InFindings of the association for computational linguistics: ACL 2024. 3563–3578

2024
[4]

Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, Dasha Metropolitan- sky, Robert Osazuwa Ness, and Jonathan Larson. 2024. From local to global: A graph rag approach to query-focused summarization.arXiv preprint arXiv:2404.16130(2024)

work page internal anchor Pith review arXiv 2024
[5]

Angela Fan, Yacine Jernite, Ethan Perez, David Grangier, Jason Weston, and Michael Auli. 2019. ELI5: Long form question answering. InProceedings of the 57th annual meeting of the association for computational linguistics. 3558–3567

2019
[6]

Luyu Gao, Zhuyun Dai, Panupong Pasupat, Anthony Chen, Arun Tejasvi Chaganty, Yicheng Fan, Vincent Zhao, Ni Lao, Hongrae Lee, Da-Cheng Juan, et al. 2023. Rarr: Researching and revising what language models say, using language models. InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 16477–16508

2023
[7]

Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yixin Dai, Jiawei Sun, Haofen Wang, Haofen Wang, et al. 2023. Retrieval-augmented generation for large language models: A survey.arXiv preprint arXiv:2312.10997 2, 1 (2023), 32

work page internal anchor Pith review Pith/arXiv arXiv 2023
[8]

Ziyu Ge, Yuhao Wu, Daniel Wai Kit Chin, Roy Ka-Wei Lee, and Rui Cao. 2025. Resolving conflicting evidence in automated fact-checking: A study on retrieval-augmented llms.arXiv preprint arXiv:2505.17762(2025)

work page arXiv 2025
[9]

Zhibin Gou, Zhihong Shao, Yeyun Gong, Yelong Shen, Yujiu Yang, Nan Duan, and Weizhu Chen. 2023. Critic: Large language models can self-correct with tool-interactive critiquing.arXiv preprint arXiv:2305.11738(2023)

work page internal anchor Pith review arXiv 2023
[10]

Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al . 2024. The llama 3 herd of models.arXiv preprint arXiv:2407.21783(2024)

work page internal anchor Pith review Pith/arXiv arXiv 2024
[11]

Bernal Jiménez Gutiérrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga, and Yu Su. 2024. HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models. InAdvances in Neural Information Processing Systems, A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang (Eds.), Vol. 37. Curran Associates, Inc., 59532–59569

2024
[12]

Rujun Han, Yuhao Zhang, Peng Qi, Yumo Xu, Jenyuan Wang, Lan Liu, William Yang Wang, Bonan Min, and Vittorio Castelli. 2024. Rag-qa arena: Evaluating domain robustness for long-form retrieval augmented question answering. In Proceedings of the 2024 conference on empirical methods in natural language processing. 4354–4374

2024
[13]

Yunhai Hu, Yilun Zhao, Chen Zhao, and Arman Cohan. 2025. Mcts-rag: Enhancing retrieval-augmented generation with monte carlo tree search.arXiv preprint arXiv:2503.20757(2025)

work page arXiv 2025
[14]

Yukun Huang, Yixin Liu, Raghuveer Thirukovalluru, Arman Cohan, and Bhuwan Dhingra. 2024. Calibrating long-form generations from large language models. InFindings of the association for computational linguistics: EMNLP 2024. 13441–13460

2024
[15]

Jeongyeon Hwang, Junyoung Park, Hyejin Park, Dongwoo Kim, Sangdon Park, and Jungseul Ok. 2025. Retrieval- augmented generation with estimation of source reliability. InProceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 34267–34291

2025
[16]

Zhan Peng Lee, Andre Lin, and Calvin Tan. 2025. Finetune-RAG: Fine-Tuning Language Models to Resist Hallucination in Retrieval-Augmented Generation.arXiv preprint arXiv:2505.10792(2025)

work page arXiv 2025
[17]

Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. InAdvances in Neural Information Processing Systems, H. Larochelle, M. Ranzato, R. Hadsell, M.F...

2020
[18]

Changmao Li and Jeffrey Flanigan. 2024. Rac: Efficient llm factuality correction with retrieval augmentation.arXiv preprint arXiv:2410.15667(2024)

work page arXiv 2024
[19]

Nelson F Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang. 2024. Lost in the middle: How language models use long contexts.Transactions of the association for computational linguistics 12 (2024), 157–173

2024
[20]

Shuyi Liu, Yu-Ming Shang, and Xi Zhang. 2026. Truthfulrag: Resolving factual-level conflicts in retrieval-augmented generation with knowledge graphs. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 40. 32168–32176

2026
[21]

Linhao Luo, Yuan-Fang Li, Gholamreza Haffari, and Shirui Pan. 2023. Reasoning on graphs: Faithful and interpretable large language model reasoning.arXiv preprint arXiv:2310.01061(2023)

work page arXiv 2023
[22]

Jacob Menick, Maja Trebacz, Vladimir Mikulik, John Aslanides, Francis Song, Martin Chadwick, Mia Glaese, Susannah Young, Lucy Campbell-Gillingham, Geoffrey Irving, et al. 2022. Teaching language models to support answers with verified quotes.arXiv preprint arXiv:2203.11147(2022)

work page arXiv 2022
[23]

Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, et al. 2021. WebGPT: Browser-Assisted Question-Answering with Human Feedback.arXiv preprint arXiv:2112.09332(2021)

work page internal anchor Pith review arXiv 2021
[24]

Ronak Pradeep, Nandan Thakur, Shivani Upadhyay, Daniel Campos, Nick Craswell, Ian Soboroff, Hoa Trang Dang, and Jimmy Lin. 2025. The great nugget recall: Automating fact extraction and rag evaluation with large language models. InProceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 180–190

2025
[25]

Yujia Qin, Zihan Cai, Dian Jin, Lan Yan, Shihao Liang, Kunlun Zhu, Yankai Lin, Xu Han, Ning Ding, Huadong Wang, et al. 2023. Webcpm: Interactive web search for chinese long-form question answering. InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 8968–8988

2023
[26]

Yingqi Qu, Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Wayne Xin Zhao, Daxiang Dong, Hua Wu, and Haifeng Wang
[27]

InProceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies

RocketQA: An Optimized Training Approach to Dense Passage Retrieval for Open-Domain Question Answering. InProceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 5835–5847

2021
[28]

Mayank Ravishankara. 2025. FVA-RAG: Falsification-Verification Alignment for Mitigating Sycophantic Hallucinations. arXiv preprint arXiv:2512.07015(2025)

work page arXiv 2025
[29]

Ruiyang Ren, Yuhao Wang, Junyi Li, Jinhao Jiang, Wayne Xin Zhao, Wenjie Wang, and Tat-Seng Chua. 2025. LLM- based Search Assistant with Holistically Guided MCTS for Intricate Information Seeking. InProceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1098–1108

2025
[30]

Ruiyang Ren, Yuhao Wang, Yingqi Qu, Wayne Xin Zhao, Jing Liu, Hua Wu, Ji-Rong Wen, and Haifeng Wang. 2025. Investigating the factual knowledge boundary of large language models with retrieval augmentation. InProceedings of the 31st International Conference on Computational Linguistics. 3697–3715

2025
[31]

Dongyu Ru, Lin Qiu, Xiangkun Hu, Tianhang Zhang, Peng Shi, Shuaichen Chang, Cheng Jiayang, Cunxiang Wang, Shichao Sun, Huanyu Li, Zizhao Zhang, Binjie Wang, Jiarong Jiang, Tong He, Zhiguo Wang, Pengfei Liu, Yue Zhang, and Zheng Zhang. 2024. RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented Generation. In Advances in Neural Informatio...

work page doi:10.52202/079017-0692 2024
[32]

Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. Reflexion: language agents with verbal reinforcement learning. InAdvances in Neural Information Processing Systems, A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine (Eds.), Vol. 36. Curran Associates, Inc., 8634–8652

2023
[33]

Jiashuo Sun, Chengjin Xu, Lumingyuan Tang, Saizhuo Wang, Chen Lin, Yeyun Gong, Lionel M Ni, Heung-Yeung Shum, and Jian Guo. 2023. Think-on-graph: Deep and responsible reasoning of large language model on knowledge graph. arXiv preprint arXiv:2307.07697(2023)

work page arXiv 2023
[34]

Karthik Valmeekam, Matthew Marquez, Alberto Olmo, Sarath Sreedharan, and Subbarao Kambhampati. 2023. Plan- Bench: An Extensible Benchmark for Evaluating Large Language Models on Planning and Reasoning about Change. InAdvances in Neural Information Processing Systems, A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine (Eds.), Vol. 36. Curr...

2023
[35]

Yuhao Wang, Ruiyang Ren, Yucheng Wang, Jing Liu, Xin Zhao, Hua Wu, and Haifeng Wang. 2026. Bee-rag: Balanced entropy engineering for retrieval-augmented generation. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 40. 33737–33745

2026
[36]

Yuhao Wang, Ruiyang Ren, Yucheng Wang, Wayne Xin Zhao, Jing Liu, Hua Wu, and Haifeng Wang. 2025. Reinforced Informativeness Optimization for Long-Form Retrieval-Augmented Generation.arXiv preprint arXiv:2505.20825(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025
[37]

Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed Chi, Quoc V Le, and Denny Zhou
[38]

InAdvances in Neural Information Processing Systems, S

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. InAdvances in Neural Information Processing Systems, S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Eds.), Vol. 35. Curran Associates, ACM Trans. Inf. Syst., Vol. 1, No. 1, Article . Publication date: April 2026. ArbGraph: Conflict-Aware Evidence Arbitration for Rel...

2026
[39]

Jerry Wei, Chengrun Yang, Xinying Song, Yifeng Lu, Nathan Hu, Jie Huang, Dustin Tran, Daiyi Peng, Ruibo Liu, Da Huang, Cosmo Du, and Quoc V. Le. 2024. Long-form factuality in large language models. InAdvances in Neural Information Processing Systems, A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang (Eds.), Vol. 37. Curran ...

work page doi:10.52202/079017-2567 2024
[40]

Zhuohan Xie, Rui Xing, Yuxia Wang, Jiahui Geng, Hasan Iqbal, Dhruv Sahnan, Iryna Gurevych, and Preslav Nakov
[41]

InFindings of the Association for Computational Linguistics: NAACL 2025

FIRE: fact-checking with iterative retrieval and verification. InFindings of the Association for Computational Linguistics: NAACL 2025. 2901–2914

2025
[42]

Shi-Qi Yan, Jia-Chen Gu, Yun Zhu, and Zhen-Hua Ling. 2024. Corrective Retrieval Augmented Generation. arXiv:2401.15884

work page internal anchor Pith review arXiv 2024
[43]

An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. 2025. Qwen3 technical report.arXiv preprint arXiv:2505.09388(2025)

work page internal anchor Pith review Pith/arXiv arXiv 2025
[44]

Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2023. React: Syner- gizing reasoning and acting in language models. InThe eleventh international conference on learning representations

2023
[45]

Ori Yoran, Tomer Wolfson, Ori Ram, and Jonathan Berant. 2023. Making retrieval-augmented language models robust to irrelevant context.arXiv preprint arXiv:2310.01558(2023)

work page arXiv 2023
[46]

Wenhao Yu, Hongming Zhang, Xiaoman Pan, Peixin Cao, Kaixin Ma, Jian Li, Hongwei Wang, and Dong Yu. 2024. Chain-of-note: Enhancing robustness in retrieval-augmented language models. InProceedings of the 2024 conference on empirical methods in natural language processing. 14672–14685

2024
[47]

Zhenrui Yue, Huimin Zeng, Yimeng Lu, Lanyu Shang, Yang Zhang, and Dong Wang. 2024. Evidence-driven retrieval augmented response generation for online misinformation. InProceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 5628–5643

2024
[48]

Xinran Zhao, Aakanksha Naik, Jay DeYoung, Joseph Chee Chang, Jena D Hwang, Tongshuang Wu, and Varsha Kishore
[49]

Improving Attributed Long-form Question Answering with Intent Awareness.arXiv preprint arXiv:2603.27435 (2026)

work page arXiv 2026
[50]

Yilun Zhao, Lyuhao Chen, Arman Cohan, and Chen Zhao. 2024. TaPERA: Enhancing faithfulness and interpretability in long-form table QA by content planning and execution-based reasoning. InProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 12824–12840

2024
[51]

Yuqicheng Zhu, Nico Potyka, Daniel Hernández, Yuan He, Zifeng Ding, Bo Xiong, Dongzhuoran Zhou, Evgeny Kharlamov, and Steffen Staab. 2025. Argrag: Explainable retrieval augmented generation using quantitative bipolar argumentation.arXiv preprint arXiv:2508.20131(2025). ACM Trans. Inf. Syst., Vol. 1, No. 1, Article . Publication date: April 2026

work page arXiv 2025