The reviewed record of science sign in
Pith

arxiv: 2606.13171 · v1 · pith:KVQCONYO · submitted 2026-06-11 · cs.CL · cs.AI

NTS-CoT: Mitigating Hallucinations in LLM-based News Timeline Summarization with Chain-of-Thought Reasoning

Feng Lyu , Huiqin Yan , Sijing Duan , Hao Wu , Shuang Gu , Xue Qiao , Weixu Zhang , Haolun Wu This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-06-27 06:53 UTCgrok-4.3pith:KVQCONYOrecord.jsonopen to challenge →

classification cs.CL cs.AI
keywords hallucinationstimeline summarizationchain-of-thoughtlarge language modelsnews summarizationdate-event pairstemporal reasoningcausal inference
0
0 comments X

The pith

NTS-CoT applies three chain-of-thought modules to cut unfaithful content and date-event omissions in LLM news timeline summarization.

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

The paper targets hallucinations in large language model timeline summarization of news, where outputs stray from source material. It distinguishes two main failure modes: unfaithful summaries of individual articles and missing information when pairing dates with events. NTS-CoT introduces targeted reasoning steps to capture key elements, select salient timestamps, and infer causal links between events. If these steps work, the resulting timelines stay closer to the original reporting while covering event sequences more completely. The authors test this on standard benchmarks with both automatic metrics and human judgments.

Core claim

NTS-CoT mitigates hallucinations through Element-CoT for faithful news element capture, Date Selection that weighs temporal saliency against event prominence, and Causal-CoT that infers relationships to avoid omissions in date-event pairs, leading to better performance than prior baselines on three TLS benchmarks.

What carries the argument

NTS-CoT framework with Element-CoT, Date Selection, and Causal-CoT modules that apply chain-of-thought reasoning to enforce faithfulness and completeness.

If this is right

  • Quantitative scores improve across the three TLS benchmarks relative to prior methods.
  • Unfaithful content decreases during the summarization stage.
  • Omissions decrease when constructing date-event pairs.
  • Human evaluators rate the outputs as more faithful and complete than baseline LLM timelines.

Where Pith is reading between the lines

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

  • The same module structure could be adapted to other sequence-generation tasks that require both factual grounding and temporal ordering.
  • Explicit causal inference steps may help reduce drift in longer multi-document summaries even outside news.
  • Date-selection logic based on combined saliency and prominence offers a template for timestamp-aware generation in domains such as legal or medical event tracking.

Load-bearing premise

The two hallucination types identified are the main problems and the three modules fix them without introducing new errors or depending on particular model behaviors.

What would settle it

A re-run of the three benchmark experiments in which NTS-CoT produces comparable or higher rates of unfaithful content and date-event omissions than the strongest baseline, or shows no gain in human faithfulness ratings.

Figures

Figures reproduced from arXiv: 2606.13171 by Feng Lyu, Haolun Wu, Hao Wu, Huiqin Yan, Shuang Gu, Sijing Duan, Weixu Zhang, Xue Qiao.

Figure 1
Figure 1. Figure 1: Examples of LLMs’ hallucinations in the news timeline summarization task. To mitigate hallucinations in LLM-based news timeline summarization, Chain-of-Thought (CoT) is an effective approach. CoT enhances the model’s ability to generate faithful and co￾herent summaries by breaking complex tasks into intermediate reasoning steps, thereby reduc￾ing errors and improving alignment with source content [19, 20, … view at source ↗
Figure 2
Figure 2. Figure 2: An overview of the NTS-CoT pipeline for TLS and an example. Given news articles and a topic query, the pipeline (1) summarizes each article with Element-CoT, (2) selects dates for the timeline, and (3) summarizes event clusters for those dates with Causal-CoT. Gray squares and green circles represent the news summaries and event clusters of the same events, respectively, with each cluster containing multip… view at source ↗
Figure 3
Figure 3. Figure 3: Parameter analysis results. At the sub-component level, we perform fine-grained ablations on Element-CoT and Causal-CoT. For Element-CoT, we sequentially remove each element, including event, entity, location, date, and result, and observe performance declines in all cases. Specifically, removing date leads to a significant drop in Date F1, highlighting the crucial role of date elements in maintaining temp… view at source ↗
Figure 4
Figure 4. Figure 4: Prompts used in Element-CoT. ### Instruction [4 guiding questions] Review the given news article associated with the provided keyword. Please answer the above questions, follow the given format strictly. ### Format [Format] ### Keyword {keyword} ### News Content {content} ### Answer ### Instruction ..., follow the given [Rules] and summarize the events related to the keyword. [Rules] ### Format [Format] ##… view at source ↗
Figure 5
Figure 5. Figure 5: Prompts used in Causal-CoT [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
read the original abstract

The rapid updates of online news make tracking event developments challenging, highlighting the need for timeline summarization (TLS). Hallucinations, where LLM-generated content deviates from source news, still remain a critical issue in LLM-based TLS and are not well studied in existing works. To bridge this gap, we identify two primary types of hallucinations: unfaithful content during news summarization and information omission in date-event summarization. Then, we propose NTS-CoT, a novel framework that leverages Chain-of-Thought (CoT) reasoning to mitigate hallucinations in TLS. The framework consists of three key modules: i) Element-CoT to capture essential news elements for faithful summarization, ii) Date Selection to combine temporal saliency and event prominence for timestamp selection, and iii) Causal-CoT to infer causal relationships and reduce omissions in date-event summarization. Extensive experiments, including quantitative analysis on three TLS benchmarks and human evaluation, demonstrate that NTS-CoT outperforms state-of-the-art baselines, effectively mitigating hallucinations and improving LLM-based TLS performance. Our source code is available at https://anonymous.4open.science/r/NTS-CoT .

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.

Referee Report

2 major / 2 minor

Summary. The manuscript presents NTS-CoT, a framework to mitigate hallucinations in LLM-based news timeline summarization (TLS). It identifies two hallucination types—unfaithful content during summarization and omissions in date-event pairs—and proposes three CoT-based modules: Element-CoT for capturing essential news elements, Date Selection combining temporal saliency and event prominence, and Causal-CoT for inferring causal relationships. The central claim is that extensive quantitative experiments on three TLS benchmarks plus human evaluation show NTS-CoT outperforms state-of-the-art baselines while reducing the identified hallucinations; source code is released.

Significance. If the empirical results and robustness claims hold, the work would be a useful contribution to LLM summarization research by offering a modular, reasoning-based approach to a practical task (news timeline construction) where hallucinations are known to be problematic. The explicit decomposition into two hallucination types and the public code release are positive for reproducibility. Significance is tempered by the absence of cross-model validation in the reported evaluation.

major comments (2)
  1. [Abstract and Experiments] Abstract and Experiments: the claim that the three modules 'effectively mitigate hallucinations' for LLM-based TLS rests on the untested premise that Element-CoT, Date Selection, and Causal-CoT succeed at element extraction and causal inference without introducing new reasoning errors; no ablation or error-propagation analysis of the intermediate CoT steps is described.
  2. [Experiments] Experiments: no results are reported across different LLM families or base-model sizes, leaving open whether the observed gains depend on the reasoning fidelity of the specific models tested rather than constituting a general mitigation strategy.
minor comments (2)
  1. [Abstract] The anonymous code link is appropriate for review but should be replaced with a permanent repository upon acceptance.
  2. [Abstract] Baseline names, exact metrics (e.g., ROUGE, hallucination rate), and dataset statistics are referenced but not enumerated in the abstract; adding a compact results table in the abstract or introduction would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive comments. We address each major comment below and indicate where revisions will be made.

read point-by-point responses

  1. Referee: [Abstract and Experiments] Abstract and Experiments: the claim that the three modules 'effectively mitigate hallucinations' for LLM-based TLS rests on the untested premise that Element-CoT, Date Selection, and Causal-CoT succeed at element extraction and causal inference without introducing new reasoning errors; no ablation or error-propagation analysis of the intermediate CoT steps is described.

    Authors: The manuscript's central evidence consists of end-to-end gains on three TLS benchmarks together with human evaluation showing reduced unfaithful content and omissions. We agree, however, that direct validation of the intermediate CoT outputs would strengthen the causal link between the modules and hallucination reduction. In the revision we will add module-level ablations and an error analysis of the Element-CoT, Date Selection, and Causal-CoT steps against reference annotations. revision: yes

  2. Referee: [Experiments] Experiments: no results are reported across different LLM families or base-model sizes, leaving open whether the observed gains depend on the reasoning fidelity of the specific models tested rather than constituting a general mitigation strategy.

    Authors: All reported results use the LLMs and settings described in the experimental section. To address generalizability, we will expand the evaluation in the revised manuscript to include at least one additional model family and a smaller base model, reporting the same metrics so that readers can assess whether the observed improvements hold across different reasoning capabilities. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical framework with benchmark evaluation

full rationale

The paper introduces NTS-CoT as an LLM-based framework with three CoT modules for TLS hallucination mitigation and supports its claims solely through quantitative experiments on three TLS benchmarks plus human evaluation. No equations, fitted parameters, derivations, or self-citation chains appear in the provided text; the central claims rest on direct performance comparisons rather than any reduction to inputs by construction. This is a standard empirical contribution with no load-bearing self-referential steps.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on standard prompting assumptions rather than new mathematical axioms or fitted constants. No free parameters, invented physical entities, or ad-hoc lemmas are introduced.

axioms (1)
  • domain assumption LLMs can follow structured chain-of-thought prompts to extract elements, select dates, and infer causal links when given appropriate instructions.
    Implicit in the design of Element-CoT, Date Selection, and Causal-CoT modules.

pith-pipeline@v0.9.1-grok · 5757 in / 1166 out tokens · 15107 ms · 2026-06-27T06:53:43.006427+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

57 extracted references · 6 canonical work pages · 3 internal anchors

  1. [1]

    Predicting relevant news events for timeline summaries,

    G. Binh Tran, M. Alrifai, and D. Quoc Nguyen, “Predicting relevant news events for timeline summaries,” inProceedings of the 22nd International Conference on World Wide Web, ser. WWW ’13 Companion. New York, NY , USA: Association for Computing Machinery, 2013, p. 91–92

    2013

  2. [2]

    Joint graphical models for date selection in timeline summarization,

    G. Tran, E. Herder, and K. Markert, “Joint graphical models for date selection in timeline summarization,” inProceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). Beijing, China: Association for Computational Linguistics, Ju...

    2015

  3. [3]

    Examining the state-of-the-art in news timeline summarization,

    D. Gholipour Ghalandari and G. Ifrim, “Examining the state-of-the-art in news timeline summarization,” inProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, D. Jurafsky, J. Chai, N. Schluter, and J. Tetreault, Eds. Online: Association for Computational Linguistics, Jul. 2020, pp. 1322–1334

    2020

  4. [4]

    Query based event extraction along a timeline,

    H. L. Chieu and Y . K. Lee, “Query based event extraction along a timeline,” inProceedings of the 27th annual international ACM SIGIR conference on Research and development in information retrieval, 2004, pp. 425–432

    2004

  5. [5]

    A temporally sensitive submodularity framework for timeline summarization,

    S. Martschat and K. Markert, “A temporally sensitive submodularity framework for timeline summarization,” inProceedings of the 22nd Conference on Computational Natural Language Learning, A. Korhonen and I. Titov, Eds. Brussels, Belgium: Association for Computational Linguistics, Oct. 2018, pp. 230–240

    2018

  6. [6]

    Language models are few-shot learners,

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert-V oss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amodei...

    2020

  7. [7]

    Extractive summarization via chatgpt for faithful summary generation,

    H. Zhang, X. Liu, and J. Zhang, “Extractive summarization via chatgpt for faithful summary generation,” inFindings of the association for computational linguistics: EMNLP 2023, 2023, pp. 3270–3278

    2023

  8. [8]

    The unreliability of explanations in few-shot prompting for textual reasoning,

    X. Ye and G. Durrett, “The unreliability of explanations in few-shot prompting for textual reasoning,” in Advances in Neural Information Processing Systems, S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh, Eds., vol. 35. Curran Associates, Inc., 2022, pp. 30 378–30 392

    2022

  9. [9]

    Evaluating the factual consistency of large language models through news summarization,

    D. Tam, A. Mascarenhas, S. Zhang, S. Kwan, M. Bansal, and C. Raffel, “Evaluating the factual consistency of large language models through news summarization,” inFindings of the Association for Computational Linguistics: ACL 2023, A. Rogers, J. Boyd-Graber, and N. Okazaki, Eds. Toronto, Canada: Association for Computational Linguistics, Jul. 2023, pp. 5220–5255

    2023

  10. [10]

    Survey of hallucination in natural language generation,

    Z. Ji, N. Lee, R. Frieske, T. Yu, D. Su, Y . Xu, E. Ishii, Y . J. Bang, A. Madotto, and P. Fung, “Survey of hallucination in natural language generation,”ACM Computing Surveys, vol. 55, no. 12, pp. 1–38, 2023

    2023

  11. [11]

    Methods for measuring, updating, and visualizing factual beliefs in language models,

    P. Hase, M. Diab, A. Celikyilmaz, X. Li, Z. Kozareva, V . Stoyanov, M. Bansal, and S. Iyer, “Methods for measuring, updating, and visualizing factual beliefs in language models,” inProceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics. Dubrovnik, Croatia: Association for Computational Linguistics, May ...

    2023

  12. [12]

    Assessing support for the trec 2024 rag track: A large-scale comparative study of llm and human evalua- tions,

    N. Thakur, R. Pradeep, S. Upadhyay, D. Campos, N. Craswell, I. Soboroff, H. T. Dang, and J. Lin, “Assessing support for the trec 2024 rag track: A large-scale comparative study of llm and human evalua- tions,” inProceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2025, pp. 2759–2763

    2024

  13. [13]

    From moments to milestones: Incremental timeline summarization leveraging large language models,

    Q. Hu, G. Moon, and H. T. Ng, “From moments to milestones: Incremental timeline summarization leveraging large language models,” inProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), L.-W. Ku, A. Martins, and V . Srikumar, Eds. Bangkok, Thailand: Association for Computational Linguistics, Aug. 2...

    2024

  14. [14]

    Web news timeline generation with extended task prompting,

    S. Wang, Y . Li, H. Xiao, L. Deng, and Y . Dong, “Web news timeline generation with extended task prompting,”arXiv preprint arXiv:2311.11652, 2023

    work page arXiv 2023

  15. [15]

    On faithfulness and factuality in abstractive sum- marization,

    J. Maynez, S. Narayan, B. Bohnet, and R. McDonald, “On faithfulness and factuality in abstractive sum- marization,” inProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, D. Jurafsky, J. Chai, N. Schluter, and J. Tetreault, Eds. Online: Association for Computational Linguistics, Jul. 2020, pp. 1906–1919

    2020

  16. [16]

    Named entity inclusion in abstractive text summarization,

    S. Berezin and T. Batura, “Named entity inclusion in abstractive text summarization,” inProceedings of the Third Workshop on Scholarly Document Processing, 2022, pp. 158–162

    2022

  17. [17]

    A Comprehensive Survey of Hallucination Mitigation Techniques in Large Language Models

    S. Tonmoy, S. Zaman, V . Jain, A. Rani, V . Rawte, A. Chadha, and A. Das, “A comprehensive survey of hallucination mitigation techniques in large language models,”arXiv preprint arXiv:2401.01313, vol. 6, 2024. 10

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  18. [18]

    A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions,

    L. Huang, W. Yu, W. Ma, W. Zhong, Z. Feng, H. Wang, Q. Chen, W. Peng, X. Feng, B. Qinet al., “A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions,” ACM Transactions on Information Systems, vol. 43, no. 2, pp. 1–55, 2025

    2025

  19. [19]

    Chain- of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, b. ichter, F. Xia, E. Chi, Q. V . Le, and D. Zhou, “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, Inc., 2022, pp. 24 824–24 837

    2022

  20. [20]

    Tree of thoughts: Deliberate problem solving with large language models,

    S. Yao, D. Yu, J. Zhao, I. Shafran, T. Griffiths, Y . Cao, and K. Narasimhan, “Tree of thoughts: Deliberate problem solving with large language models,” 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., 2023, pp. 11 809–11 822

    2023

  21. [21]

    Element-aware summarization with large language models: Expert- aligned evaluation and chain-of-thought method,

    Y . Wang, Z. Zhang, and R. Wang, “Element-aware summarization with large language models: Expert- aligned evaluation and chain-of-thought method,” inProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), A. Rogers, J. Boyd-Graber, and N. Okazaki, Eds. Toronto, Canada: Association for Computational ...

    2023

  22. [22]

    Leveraging learning to rank in an optimization framework for timeline summarization,

    G. B. Tran, T. A. Tran, N.-K. Tran, M. Alrifai, and N. Kanhabua, “Leveraging learning to rank in an optimization framework for timeline summarization,” inSIGIR 2013 Workshop on Time-aware Information Access (TAIA, 2013

    2013

  23. [23]

    Timeline summarization from relevant headlines,

    G. Tran, M. Alrifai, and E. Herder, “Timeline summarization from relevant headlines,” inAdvances in Information Retrieval: 37th European Conference on IR Research, ECIR 2015, Vienna, Austria, March 29-April 2, 2015. Proceedings 37. Springer, 2015, pp. 245–256

    2015

  24. [24]

    The structure and function of communication in society,

    H. D. Lasswell, “The structure and function of communication in society,”The communication of ideas, vol. 37, no. 1, pp. 136–139, 1948

    1948

  25. [25]

    Improving faithfulness of large language models in summarization via sliding generation and self-consistency,

    T. Li, Z. Li, and Y . Zhang, “Improving faithfulness of large language models in summarization via sliding generation and self-consistency,” inProceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), 2024, pp. 8804–8817

    2024

  26. [26]

    Multilingual and cross-domain temporal tagging,

    J. Strötgen and M. Gertz, “Multilingual and cross-domain temporal tagging,”Language Resources and Evaluation, vol. 47, pp. 269–298, 2013

    2013

  27. [27]

    Summarize dates first: A paradigm shift in timeline summarization,

    M. La Quatra, L. Cagliero, E. Baralis, A. Messina, and M. Montagnuolo, “Summarize dates first: A paradigm shift in timeline summarization,” inProceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2021, pp. 418–427

    2021

  28. [28]

    Timeline summarization based on event graph compression via time-aware optimal transport,

    M. Li, T. Ma, M. Yu, L. Wu, T. Gao, H. Ji, and K. McKeown, “Timeline summarization based on event graph compression via time-aware optimal transport,” inProceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, M.-F. Moens, X. Huang, L. Specia, and S. W.-t. Yih, Eds. Online and Punta Cana, Dominican Republic: Association for ...

    2021

  29. [29]

    Abstractive timeline summarization,

    J. Steen and K. Markert, “Abstractive timeline summarization,” inProceedings of the 2nd Workshop on New Frontiers in Summarization, L. Wang, J. C. K. Cheung, G. Carenini, and F. Liu, Eds. Hong Kong, China: Association for Computational Linguistics, Nov. 2019, pp. 21–31

    2019

  30. [30]

    The Llama 3 Herd of Models

    A. Dubey, A. Jauhri, A. Pandey, A. Kadian, A. Al-Dahle, A. Letman, A. Mathur, A. Schelten, A. Yang, A. Fanet al., “The llama 3 herd of models,”arXiv preprint arXiv:2407.21783, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  31. [31]

    Automatic generation of overview timelines,

    R. Swan and J. Allan, “Automatic generation of overview timelines,” inProceedings of the 23rd annual international ACM SIGIR conference on Research and development in information retrieval, 2000, pp. 49–56

    2000

  32. [32]

    Towards timeline generation with abstract meaning representa- tion,

    B. Mansouri, R. Campos, and A. Jatowt, “Towards timeline generation with abstract meaning representa- tion,” inCompanion Proceedings of the ACM Web Conference 2023, 2023, pp. 1204–1207

    2023

  33. [33]

    Timeline generation through evolutionary trans-temporal summarization,

    R. Yan, L. Kong, C. Huang, X. Wan, X. Li, and Y . Zhang, “Timeline generation through evolutionary trans-temporal summarization,” inProceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, 2011, pp. 433–443

    2011

  34. [34]

    Learning towards abstractive timeline summa- rization,

    X. Chen, Z. Chan, S. Gao, M.-H. Yu, D. Zhao, and R. Yan, “Learning towards abstractive timeline summa- rization,” inProceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19. International Joint Conferences on Artificial Intelligence Organization, 7 2019, pp. 4939–4945

    2019

  35. [35]

    Follow the timeline! generating an abstractive and extractive timeline summary in chronological order,

    X. Chen, M. Li, S. Gao, Z. Chan, D. Zhao, X. Gao, X. Zhang, and R. Yan, “Follow the timeline! generating an abstractive and extractive timeline summary in chronological order,”ACM Transactions on Information Systems, vol. 41, no. 1, pp. 1–30, 2023

    2023

  36. [36]

    Timeline summarization in the era of llms,

    D. Sojitra, R. Jain, S. Saha, A. Jatowt, and M. Gupta, “Timeline summarization in the era of llms,” in Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2024, pp. 2657–2661. 11

    2024

  37. [37]

    Combining hierachical V AEs with LLMs for clinically meaningful timeline summarisation in social media,

    J. Song, J. Chim, A. Tsakalidis, J. Ive, D. Atzil-Slonim, and M. Liakata, “Combining hierachical V AEs with LLMs for clinically meaningful timeline summarisation in social media,” inFindings of the Association for Computational Linguistics: ACL 2024, L.-W. Ku, A. Martins, and V . Srikumar, Eds. Bangkok, Thailand: Association for Computational Linguistics,...

    2024

  38. [38]

    Unfolding the headline: Iterative self- questioning for news retrieval and timeline summarization,

    W. Wu, S. Huang, Y . Jiang, P. Xie, F. Huang, and H. Zhao, “Unfolding the headline: Iterative self- questioning for news retrieval and timeline summarization,” inFindings of the Association for Computa- tional Linguistics: NAACL 2025, 2025, pp. 4385–4398

    2025

  39. [39]

    Just what you desire: Constrained timeline summarization with self-reflection for enhanced relevance,

    M. R. Qorib, Q. Hu, and H. T. Ng, “Just what you desire: Constrained timeline summarization with self-reflection for enhanced relevance,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 39, no. 23, 2025, pp. 25 065–25 073

    2025

  40. [40]

    “why is this misleading?

    J. Shen, J. Liu, D. Finnie, N. Rahmati, M. Bendersky, and M. Najork, ““why is this misleading?”: Detecting news headline hallucinations with explanations,” inProceedings of the ACM Web Conference 2023, 2023, pp. 1662–1672

    2023

  41. [41]

    AlignSum: Data pyramid hierarchical fine-tuning for aligning with human summarization preference,

    Y . Han, Y . Wang, R. Wang, L. Chen, and K. Yu, “AlignSum: Data pyramid hierarchical fine-tuning for aligning with human summarization preference,” inFindings of the Association for Computational Linguistics: EMNLP 2024, Y . Al-Onaizan, M. Bansal, and Y .-N. Chen, Eds. Miami, Florida, USA: Association for Computational Linguistics, Nov. 2024, pp. 8506–8522

    2024

  42. [42]

    Alleviating hallucinations in large language models with scepticism modeling,

    Y . Wu, Y . Wang, T. Chen, C. Liu, N. Xi, Q. Gu, H. Lei, Z. Jiang, Y . Chen, and L. Ji, “Alleviating hallucinations in large language models with scepticism modeling,”arXiv preprint arXiv:2409.06601, 2024

    work page arXiv 2024

  43. [43]

    R-tuning: Instructing large language models to say ‘i don’t know’,

    H. Zhang, S. Diao, Y . Lin, Y . Fung, Q. Lian, X. Wang, Y . Chen, H. Ji, and T. Zhang, “R-tuning: Instructing large language models to say ‘i don’t know’,” inProceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). Mexico City, Mexico: Associatio...

    2024

  44. [44]

    Factually consistent summarization via reinforcement learning with textual entailment feedback,

    P. Roit, J. Ferret, L. Shani, R. Aharoni, G. Cideron, R. Dadashi, M. Geist, S. Girgin, L. Hussenot, O. Keller, N. Momchev, S. Ramos Garea, P. Stanczyk, N. Vieillard, O. Bachem, G. Elidan, A. Hassidim, O. Pietquin, and I. Szpektor, “Factually consistent summarization via reinforcement learning with textual entailment feedback,” inProceedings of the 61st An...

    2023

  45. [45]

    Mutual information alleviates hallucinations in abstractive summarization,

    L. van der Poel, R. Cotterell, and C. Meister, “Mutual information alleviates hallucinations in abstractive summarization,” inProceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, Y . Goldberg, Z. Kozareva, and Y . Zhang, Eds. Abu Dhabi, United Arab Emirates: Association for Computational Linguistics, Dec. 2022, pp. 5956–5965

    2022

  46. [46]

    Mitigating hallucination in abstractive summarization with domain- conditional mutual information,

    K. Chae, J. Choi, Y . Jo, and T. Kim, “Mitigating hallucination in abstractive summarization with domain- conditional mutual information,” inFindings of the Association for Computational Linguistics: NAACL 2024, K. Duh, H. Gomez, and S. Bethard, Eds. Mexico City, Mexico: Association for Computational Linguistics, Jun. 2024, pp. 1809–1820

    2024

  47. [47]

    Trusting your evidence: Hallucinate less with context-aware decoding,

    W. Shi, X. Han, M. Lewis, Y . Tsvetkov, L. Zettlemoyer, and W.-t. Yih, “Trusting your evidence: Hallucinate less with context-aware decoding,” inProceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 2: Short Papers), K. Duh, H. Gomez, and S. Bethard, Eds. Mexi...

    2024

  48. [48]

    Alleviating hallucinations of large language models through induced hallucinations,

    Y . Zhang, L. Cui, S. Shiet al., “Alleviating hallucinations of large language models through induced hallucinations,” inFindings of the Association for Computational Linguistics: NAACL 2025, 2025, pp. 8218–8232

    2025

  49. [49]

    SelfCheckGPT: Zero-resource black-box hallucination detection for generative large language models,

    P. Manakul, A. Liusie, and M. Gales, “SelfCheckGPT: Zero-resource black-box hallucination detection for generative large language models,” inProceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, H. Bouamor, J. Pino, and K. Bali, Eds. Singapore: Association for Computational Linguistics, Dec. 2023, pp. 9004–9017

    2023

  50. [50]

    Self-checker: Plug-and-play modules for fact-checking with large language models,

    M. Li, B. Peng, M. Galley, J. Gao, and Z. Zhang, “Self-checker: Plug-and-play modules for fact-checking with large language models,” inFindings of the Association for Computational Linguistics: NAACL 2024, K. Duh, H. Gomez, and S. Bethard, Eds. Mexico City, Mexico: Association for Computational Linguistics, Jun. 2024, pp. 163–181

    2024

  51. [51]

    Verify-and-edit: A knowledge-enhanced chain-of-thought framework,

    R. Zhao, X. Li, S. Joty, C. Qin, and L. Bing, “Verify-and-edit: A knowledge-enhanced chain-of-thought framework,” inProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), A. Rogers, J. Boyd-Graber, and N. Okazaki, Eds. Toronto, Canada: Association for Computational Linguistics, Jul. 2023, pp. 5823–5840. 12

    2023

  52. [52]

    Rethinking with retrieval: Faithful large language model inference,

    H. He, H. Zhang, and D. Roth, “Rethinking with retrieval: Faithful large language model inference,”arXiv preprint arXiv:2301.00303, 2022

    work page arXiv 2022

  53. [53]

    3a-cot: an attend-arrange-abstract chain-of-thought for multi-document summarization,

    Y . Zhang, S. Gao, Y . Huang, Z. Yu, and K. Tan, “3a-cot: an attend-arrange-abstract chain-of-thought for multi-document summarization,”International Journal of Machine Learning and Cybernetics, pp. 1–19, 2024

    2024

  54. [54]

    Autosumm: A comprehensive framework for llm-based conversation summarization,

    A. Gupta, D. Singh, G. A. Cowan, N. Kadhiresan, S. Srivastava, Y . Sriraja, and Y . K. Mantri, “Autosumm: A comprehensive framework for llm-based conversation summarization,” inProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track), 2025, pp. 500–509

    2025

  55. [55]

    Efficient memory management for large language model serving with pagedattention,

    W. Kwon, Z. Li, S. Zhuang, Y . Sheng, L. Zheng, C. H. Yu, J. Gonzalez, H. Zhang, and I. Stoica, “Efficient memory management for large language model serving with pagedattention,” inProceedings of the 29th Symposium on Operating Systems Principles, ser. SOSP ’23. New York, NY , USA: Association for Computing Machinery, 2023, p. 611–626

    2023

  56. [56]

    Towards General Text Embeddings with Multi-stage Contrastive Learning

    Z. Li, X. Zhang, Y . Zhang, D. Long, P. Xie, and M. Zhang, “Towards general text embeddings with multi-stage contrastive learning,”arXiv preprint arXiv:2308.03281, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  57. [57]

    Improving ROUGE for timeline summarization,

    S. Martschat and K. Markert, “Improving ROUGE for timeline summarization,” inProceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers. Valencia, Spain: Association for Computational Linguistics, Apr. 2017, pp. 285–290. A Discussions A.1 Limitations Although NTS-CoT has shown effect...

    2017