Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

The Truth Becomes Clearer Through Debate! Multi-Agent Systems with Large Language Models Unmask Fake News

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read Structured LLM debate improves fake news detection over single-model and multi-agent baselines.

desk verdict Worth a careful look, but the 'outperforms all baselines' claim rests on an unfair backbone comparison: ARG was run on GPT-3.5, TED on GPT-4o-mini. read the letter →

arxiv 2505.08532 v1 pith:DLQLNGCC submitted 2025-05-13 cs.SI cs.AI

classification cs.SIcs.AI
keywords fakenewsdetectionmulti-agentsystemlargelanguagemodelsdebateinterpretabilitygraphattentionnetworksocialmedia
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that fake news detection improves when multiple large language model agents debate the news in a formal structure instead of classifying it in one pass. It introduces TruEDebate (TED), in which one team argues the news is true, another argues it is fake, and a synthesis-and-analysis stage converts the debate transcript into a label. Across the ARG-EN and ARG-CN datasets, TED reports the highest macro-F1 and accuracy, including gains over the strongest prior LLM-plus-small-model method, ARG. The point of the design is not only accuracy: the debate transcript and synthesis report are meant to make the detection decision explainable to a reader.

What carries the argument

The central mechanism is the two-stage TED pipeline. DebateFlow Agents organize LLM-powered agents into Proponents and Opponents, who argue in Lincoln-Douglas style stages, producing a debate log. InsightFlow Agents then process that log: the Synthesis Agent condenses it into a report, and the Analysis Agent embeds each debate utterance with its role, builds a debate graph, applies Graph Attention Networks to model argument relationships, and uses interactive attention between the debate representation and the news text to classify. The role-aware encoding and graph structure are what let the model use who said what, not just what was said.

What would settle it

Swap the stances assigned to the two teams for the same news items and retrain or re-run the pipeline: if the predictions flip or degrade while performance on the original stance assignments stays high, the system is following assigned roles and rhetorical cues rather than the truth content of the news.

Watch

Extended reading notes

Core claim

TED's central claim is that a structured adversarial exchange surfaces signals that single-pass classifiers miss. In TED, DebateFlow Agents produce opening statements, cross-examinations, rebuttals, and closing statements from fixed pro and con stances; the Synthesis Agent then summarizes the debate around five veracity cues (specific verifiable details, credible sourcing, neutral tone, absence of emotional language, and external corroboration); and the Analysis Agent, a role-aware encoder over a debate graph, produces the final verdict. On ARG-EN, TED reaches a macro-F1 of 0.803 and accuracy of 0.892; on ARG-CN, 0.795 and 0.798, topping all reported baselines. Ablations attribute the largest drop to removing the Analysis Agent, with smaller but consistent drops from removing the DebateFlow or Synthesis Agents. The authors take these results as evidence that formal debate plus learned synthesis, rather than prompt-only LLM reasoning, is what drives detection.

Load-bearing premise

The five cues the Synthesis Agent is told to weigh—verifiable specifics, reliable sourcing, neutral tone, absence of emotional language, and external corroboration—are assumed to track whether a news item is actually fake; if they do not, the classifier may be fitting debate-style artifacts and the reported gains would not transfer.

Editorial extensions

If this is right

  • Structured debate improves on single-pass LLM classification: on both datasets, TED beats the same backbone used alone by clear margins.
  • Structured debate also beats a generic multi-agent debate scheme: TED outperforms ChatEval in both the one-by-one and simultaneous-talk configurations.
  • Each pipeline component contributes: removing the DebateFlow Agents, the Synthesis Agent, or the Analysis Agent lowers macro-F1, with the Analysis Agent's removal causing the largest drop.
  • The benefit is backbone-independent: TED improves over the LLM-only baseline whether the debate runs on GPT-4o-mini, Deepseek, Qwen 2.5, or Llama 3.1.
  • The debate transcript and synthesis report double as an explanation of the prediction, which is a stated goal and a practical advantage over black-box classifiers.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A stance-swap test, in which Proponents are told to defend 'fake' and Opponents 'true' for the same news, would show whether the model tracks assigned rhetoric rather than veracity; the paper does not run this test.
  • Because the Synthesis Agent's five cues closely match standard journalistic verification heuristics, the same pipeline could plausibly generalize to claim verification beyond news; the paper only demonstrates news classification.
  • The Analysis Agent requires fine-tuning per dataset, so a practical open question, untouched here, is whether debate logs transfer across languages or domains without retraining.
  • If adversarial argumentation is the active ingredient, pairing TED with external evidence retrieval could push accuracy further, although the paper deliberately avoids tool use.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes TruEDebate (TED), a multi-agent LLM framework for fake news detection. In TED, DebateFlow Agents simulate a Lincoln-Douglas-style debate between Proponent and Opponent teams; a Synthesis Agent summarizes the debate log; and an Analysis Agent, built from a role-aware encoder, a debate graph processed by GATs, and an interactive attention mechanism over the debate and news embeddings, produces the final veracity label. Experiments on ARG-EN and ARG-CN compare TED with LLM-only, SLM-only, and LLM+SLM baselines, and include ablations and backbone-replacement experiments.

Significance. If the headline comparison were made fair, TED would be a useful contribution to interpretable fake news detection: it combines structured multi-agent debate with a trainable graph-based analysis component, attacks an important problem, and ships code. The paper's concrete strengths are the two-dataset evaluation, the ablation evidence that each component contributes, the multiple-backbone robustness check, and the interpretable case study. The classifier is trained and evaluated on held-out splits, so there is no circularity in the technical sense. However, the central claim that TED 'outperforms all other methods' is not yet established because the strongest LLM+SLM competitor, ARG, is run with an older and likely weaker LLM backbone than TED. The significance of the contribution therefore remains conditional on a controlled comparison.

major comments (4)
  1. [Section 4.1.2, 4.1.3, Table 1, Table 4] The comparison against ARG/ARG-D is confounded by the LLM backbone. Table 1 states that all baselines except GPT-4o-mini and ChatEval are taken from Hu et al. [15], whose ARG system predates GPT-4o-mini, while TED uses GPT-4o-mini for its DebateFlow and Synthesis Agents (Section 4.1.3). The macF1 margin over ARG is only +0.013 on ARG-EN and +0.011 on ARG-CN, and Table 4 shows that TED with Llama 3.1-70B reaches exactly the ARG-reported macF1 of 0.790 on ARG-EN. The paper never re-runs ARG or ARG-D with GPT-4o-mini, nor any condition that holds the backbone fixed while removing only the debate structure. Consequently, the Section 4.2 conclusion that TED 'outperforms all other methods' conflates the debate contribution with the contribution of a stronger LLM. Please add a controlled comparison, e.g., ARG with GPT-4o-mini, or TED with the same backbone used for ARG in Hu et al.
  2. [Tables 1, 3, 4; Section 4.1.3] Statistical significance is claimed for bold and underlined entries (t-test, p<0.01), but no run count, random seed, or standard deviation is reported anywhere in the paper. A single point estimate per condition cannot support pairwise t-test claims. Please report means and standard deviations over multiple independent runs (at least 3-5 seeds) and state the exact test, sample size, and pairing used.
  3. [Section 3.4.2 and Section 4.1.3] The debate graph and the ablation variants are under-specified. Equation (7) defines GAT updates on G=(V,E), but the paper does not say how many nodes a debate produces, what edge types exist, whether edges are directed, how many GAT layers and heads are used, or how the 'sequential order or explicit references' mentioned in Section 3.4.2 is encoded. Likewise, Table 3 does not define what replaces DebateFlow Agents in 'w/o DebateFlow Agents' or what replaces the Analysis Agent in 'w/o Analysis Agent'. Without these definitions, the ablation drops cannot be interpreted and the method cannot be reproduced from the text alone.
  4. [Section 3.4.1 and Section 5.3] The interpretability claim is not validated against ground truth. The Synthesis Agent's prompt hard-codes five heuristics (verifiable details, reliable sources, neutral tone, emotional language, corroboration), and the case study merely illustrates a generated summary. The paper does not measure whether the debate winner or the Synthesis summary actually agrees with the gold label more often than chance, nor whether the Analysis Agent's performance is driven by these stylistic cues rather than by debate structure. Please add an analysis of debate-outcome/gold-label agreement, or at least explicitly frame these heuristics as assumptions and test their correlation on the validation set.
minor comments (4)
  1. [Algorithm 1] Line 18 of Algorithm 1 reads 'Build Dabate graph'; this should be 'Build Debate graph'.
  2. [Section 4.1.2 and Table 1] The baseline is listed as 'ENDEFA' in Section 4.1.2 but as 'ENDEF' in Table 1; please reconcile the name.
  3. [Abstract footnote and Appendix A] The GitHub URL spells the repository as 'TED_fake-news-detction'; this appears to be a typo for 'detection' and should be corrected.
  4. [Section 4.2 and Figure 3] The ChatEval adaptation is mentioned only as 'applying these strategies to our fake news detection task'; please specify the prompts or provide a code pointer. Also, 'Deepseek' appears with inconsistent capitalization in Figure 3 and the surrounding text.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: TED's classifier is trained on ground-truth labels with debate summaries as input features, and no prediction reduces to a fitted parameter by construction.

full rationale

The paper's derivation chain is self-contained. DebateFlow Agents generate a debate log D from the news item and fixed role prompts (Eqs. 1-5); the Synthesis Agent summarizes D into S using an explicit prompt (Section 3.4.1); the Analysis Agent encodes D and S with a role-aware encoder, GAT, and interactive attention, and is trained with cross-entropy against ground-truth labels (Eq. 12). No equation defines a target in terms of itself, and no fitted constant is relabeled as a prediction. The synthesis prompt's veracity cues (verifiable details, reliable sources, neutral tone, absence of emotional language, corroboration) are input features, not outputs, and their validity is an empirical assumption, not a circular reduction. Baseline numbers for ARG, ARG-D, and the SLM-only methods are imported from Hu et al. [15], with only GPT-4o-mini and ChatEval run by the present authors (Table 1 note, Section 4.1.2); importing an external benchmark is independent evidence, though the paper does not re-run ARG with GPT-4o-mini, so the headline comparison may be confounded by LLM backbone. That is a correctness/experimental-control concern, not a circularity. Self-citations in Related Work ([27], [28], [54]-[56]) are contextual and do not carry the central argument. The central claim is therefore not forced by definition, by a fitted quantity, or by a self-citation chain.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim is an empirical performance claim. Its main burdens are the learned Analysis Agent parameters, the chosen graph topology, and the assumption that LLM debate summaries carry veracity signal. No new physical entities are introduced; the agents are software components with internal ablation support.

free parameters (3)
  • Analysis Agent network weights (BERT fine-tune, role projection W_role, GAT layers, projections W_g and W_e… = trained on ARG-EN and ARG-CN train splits
    These learned parameters are fit to the training data and are essential to the reported classification performance. They are standard supervised-learning parameters, not ad hoc constants.
  • Hyperparameters (learning rate, grid search range, number of GAT layers, pooling choice) = not reported; chosen by grid search
    The paper states 'comprehensive grid search' but does not report the chosen values, so the fitted configuration is not recoverable from the text.
  • Debate graph edge structure = predefined based on debate sequence and role interactions; exact edges not specified
    The graph topology determines what the GAT can attend to and is a design choice that affects the Analysis Agent's predictions, but the paper does not define it precisely.
assumptions (4)
  • domain assumption Ground-truth labels in ARG-EN and ARG-CN are correct and the temporal splits used by Hu et al. prevent data leakage.
    All evaluation and training rely on these labels and splits; the paper adopts them from prior work without re-validation.
  • domain assumption LLM debate agents produce coherent, stance-consistent arguments that simulate genuine reasoning.
    The DebateFlow design in Eqs. (1)-(5) assumes GPT-4o-mini and other backbones follow the role and stance prompts; the paper does not measure prompt compliance.
  • ad hoc to paper The Synthesis Agent's five heuristics (verifiable details, reliable sources, neutral tone, emotional language, corroboration) are informative about news veracity.
    These criteria are inserted into the prompt in Section 3.4.1 and are not derived from or validated against external fact-checking standards.
  • standard math Standard backpropagation and cross-entropy training of the Analysis Agent learns a generalizable decision boundary.
    The method relies on conventional supervised training in Eq. (12); the usual i.i.d. generalization assumption is implicit.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Truth Becomes Clearer Through Debate! Multi-Agent Systems with Large Language Models Unmask Fake News." pith.science (2026). https://pith.science/paper/DLQLNGCC

@misc{pith2026250508532,
  author       = {Pith},
  title        = {Pith review of: The Truth Becomes Clearer Through Debate! Multi-Agent Systems with Large Language Models Unmask Fake News},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DLQLNGCC}},
  note         = {Machine review of arXiv:2505.08532}
}
read the original abstract

In today's digital environment, the rapid propagation of fake news via social networks poses significant social challenges. Most existing detection methods either employ traditional classification models, which suffer from low interpretability and limited generalization capabilities, or craft specific prompts for large language models (LLMs) to produce explanations and results directly, failing to leverage LLMs' reasoning abilities fully. Inspired by the saying that "truth becomes clearer through debate," our study introduces a novel multi-agent system with LLMs named TruEDebate (TED) to enhance the interpretability and effectiveness of fake news detection. TED employs a rigorous debate process inspired by formal debate settings. Central to our approach are two innovative components: the DebateFlow Agents and the InsightFlow Agents. The DebateFlow Agents organize agents into two teams, where one supports and the other challenges the truth of the news. These agents engage in opening statements, cross-examination, rebuttal, and closing statements, simulating a rigorous debate process akin to human discourse analysis, allowing for a thorough evaluation of news content. Concurrently, the InsightFlow Agents consist of two specialized sub-agents: the Synthesis Agent and the Analysis Agent. The Synthesis Agent summarizes the debates and provides an overarching viewpoint, ensuring a coherent and comprehensive evaluation. The Analysis Agent, which includes a role-aware encoder and a debate graph, integrates role embeddings and models the interactions between debate roles and arguments using an attention mechanism, providing the final judgment.

Figures

Figures reproduced from arXiv: 2505.08532 by the authors.

Figure 1
Figure 1. (a) Traditional models primarily rely on black box [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Our TED framework consists of two main components: (a) the DebateFlow Agents and (b) the InsightFlow Agents. The [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Performance of our framework TED with GPT-4o [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: A real debate case study, with its debate report, can [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SWE-Debate: Competitive Multi-Agent Debate for Software Issue Resolution

    cs.SE 2025-07 conditional novelty 6.0 of 10

    A competitive multi-agent debate over graph-derived fault propagation chains lifts automated issue resolution on SWE-bench Verified to 41.4% pass@1.

Reference graph

Works this paper leans on

67 extracted references · 32 canonical work pages · cited by 1 Pith paper

  1. [15]

    Beizhe Hu, Qiang Sheng, Juan Cao, Yuhui Shi, Yang Li, Danding Wang, and Peng Qi. 2024. Bad actor, good advisor: Exploring the role of large language models in fake news detection. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 22105–22113

  2. [1]

    Johan Bollen, Huina Mao, and Xiaojun Zeng. 2011. Twitter mood predicts the stock market.Journal of computational science2, 1 (2011), 1–8

  3. [2]

    2013.Debate and critical analysis: The harmony of conflict

    Robert James Branham. 2013.Debate and critical analysis: The harmony of conflict. Routledge

  4. [3]

    Chi-Min Chan, Weize Chen, Yusheng Su, Jianxuan Yu, Wei Xue, Shanghang Zhang, Jie Fu, and Zhiyuan Liu. 2024. ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate. InThe Twelfth International Conference on Learning Representations

  5. [4]

    Justin Chih-Yao Chen, Swarnadeep Saha, and Mohit Bansal. 2023. Reconcile: Round-table conference improves reasoning via consensus among diverse llms. arXiv preprint arXiv:2309.13007(2023)

  6. [5]

    Xiuying Chen, Mingzhe Li, Shen Gao, Xin Cheng, Qiang Yang, Qishen Zhang, Xin Gao, and Xiangliang Zhang. 2023. A Topic-aware Summarization Framework with Different Modal Side Information.SIGIR(2023)

  7. [6]

    Xiuying Chen, Guodong Long, Chongyang Tao, Mingzhe Li, Xin Gao, Chengqi Zhang, and Xiangliang Zhang. 2023. Improving the Robustness of Summarization Systems with Dual Augmentation.ACL(2023)

  8. [7]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North. doi:10.18653/v1/n19-1423

Show all 67 references
  1. [8]

    Marko Djuranovic. 2003. The Ultimate Lincoln-Douglas Debate Handbook. Retrieved Sep16 (2003), 2008

  2. [9]

    Yilun Du, Shuang Li, Antonio Torralba, Joshua B Tenenbaum, and Igor Mordatch

  3. [10]

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

  4. [11]

    Nir Grinberg, Kenneth Joseph, Lisa Friedland, Briony Swire-Thompson, and David Lazer. 2019. Fake news on Twitter during the 2016 US presidential election. Science363, 6425 (2019), 374–378

  5. [12]

    Mingfei Guo, Xiuying Chen, Juntao Li, Dongyan Zhao, and Rui Yan. 2021. How does truth evolve into fake news? an empirical study of fake news evolution. In Companion Proceedings of the Web Conference 2021. 407–411

  6. [13]

    Aditi Gupta, Hemank Lamba, Ponnurangam Kumaraguru, and Anupam Joshi

  7. [14]

    Sirui Hong, Xiawu Zheng, Jonathan Chen, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, et al . 2023. Metagpt: Meta programming for multi-agent collaborative framework.arXiv preprint arXiv:2308.00352(2023)

  8. [16]

    Beizhe Hu, Qiang Sheng, Juan Cao, Yongchun Zhu, Danding Wang, Zhengjia Wang, and Zhiwei Jin. 2023. Learn over past, evolve for future: Forecasting temporal trends for fake news detection.arXiv preprint arXiv:2306.14728(2023)

  9. [17]

    Yiqiao Jin, Xiting Wang, Ruichao Yang, Yizhou Sun, Wei Wang, Hao Liao, and Xing Xie. 2022. Towards fine-grained reasoning for fake news detection. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 36. 5746–5754

  10. [18]

    Zhao Kaiya, Michelangelo Naim, Jovana Kondic, Manuel Cortes, Jiaxin Ge, Shuy- ing Luo, Guangyu Robert Yang, and Andrew Ahn. 2023. Lyfe Agents: Generative agents for low-cost real-time social interactions.arXiv preprint arXiv:2310.02172 (2023)

  11. [19]

    Rohit Kumar Kaliyar, Anurag Goswami, and Pratik Narang. 2021. FakeBERT: Fake news detection in social media with a BERT-based deep learning approach. Multimedia tools and applications80, 8 (2021), 11765–11788

  12. [20]

    Jackie Kazil, David Masad, and Andrew Crooks. 2020. Utilizing python for agent-based modeling: The mesa framework. InSocial, Cultural, and Behavioral Modeling: 13th International Conference, SBP-BRiMS 2020, Washington, DC, USA, October 18–21, 2020, Proceedings 13. Springer, 308–317

  13. [21]

    Akbir Khan, John Hughes, Dan Valentine, Laura Ruis, Kshitij Sachan, Ansh Radhakrishnan, Edward Grefenstette, Samuel R Bowman, Tim Rocktäschel, and Ethan Perez. 2024. Debating with More Persuasive LLMs Leads to More Truthful Answers. InForty-first International Conference on Ma...

  14. [22]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic opti- mization.arXiv preprint arXiv:1412.6980(2014)

  15. [23]

    Chao Li, Xing Su, Chao Fan, Haoying Han, Cong Xue, and Chunmo Zheng. 2023. Quantifying the impact of large language models on collective opinion dynamics. arXiv preprint arXiv:2308.03313(2023)

  16. [24]

    Guohao Li, Hasan Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem. 2023. Camel: Communicative agents for" mind" exploration of large language model society.Advances in Neural Information Processing Systems36 (2023), 51991–52008

  17. [25]

    Xinyi Li, Yongfeng Zhang, and Edward C Malthouse. 2024. Large Language Model Agent for Fake News Detection.arXiv preprint arXiv:2405.01593(2024)

  18. [26]

    Tian Liang, Zhiwei He, Wenxiang Jiao, Xing Wang, Yan Wang, Rui Wang, Yujiu Yang, Shuming Shi, and Zhaopeng Tu. 2023. Encouraging divergent thinking in large language models through multi-agent debate.arXiv preprint arXiv:2305.19118(2023)

  19. [27]

    Yuhan Liu, Xiuying Chen, Xiaoqing Zhang, Xing Gao, Ji Zhang, and Rui Yan

  20. [28]

    Yuhan Liu, Zirui Song, Xiaoqing Zhang, Xiuying Chen, and Rui Yan. 2024. From a tiny slip to a giant leap: An llm-based simulation for fake news evolution.arXiv preprint arXiv:2410.19064(2024)

  21. [29]

    Yuxuan Liu, Hongda Sun, Wenya Guo, Xinyan Xiao, Cunli Mao, Zhengtao Yu, and Rui Yan. 2025. Bidev: Bilateral defusing verification for complex claim fact- checking. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 541–549

  22. [30]

    Yuxuan Liu, Hongda Sun, Wei Liu, Jian Luan, Bo Du, and Rui Yan. 2025. Mo- bileSteward: Integrating Multiple App-Oriented Agents with Self-Evolution to Automate Cross-App Instructions. InProceedings of the 31st ACM SIGKDD Con- ference on Knowledge Discovery and Data Mining V.1(...

  23. [31]

    Ahmadreza Mosallanezhad, Mansooreh Karami, Kai Shu, Michelle V Mancenido, and Huan Liu. 2022. Domain adaptive fake news detection via reinforcement learning. InProceedings of the ACM Web Conference 2022. 3632–3640

  24. [32]

    Qiong Nan, Juan Cao, Yongchun Zhu, Yanyan Wang, and Jintao Li. 2021. MD- FEND: Multi-domain fake news detection. InProceedings of the 30th ACM Inter- national Conference on Information & Knowledge Management. 3343–3347

  25. [33]

    Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. InProceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. 1–22

  26. [34]

    Joon Sung Park, Lindsay Popowski, Carrie Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2022. Social simulacra: Creating populated prototypes for social computing systems. InProceedings of the 35th Annual ACM Symposium on User Interface Software and Technol...

  27. [35]

    Chen Qian, Xin Cong, Cheng Yang, Weize Chen, Yusheng Su, Juyuan Xu, Zhiyuan Liu, and Maosong Sun. 2023. Communicative agents for software development. arXiv preprint arXiv:2307.079246 (2023)

  28. [36]

    Feng Qian, Chengyue Gong, Karishma Sharma, and Yan Liu. 2018. Neural User Response Generator: Fake News Detection with Collective User Intelligence.. In IJCAI, Vol. 18. 3834–3840

  29. [37]

    Yoel Roth. 2022. The vast majority of content we take action on for misinformation is identified proactively

  30. [38]

    Kai Shu, Deepak Mahudeswaran, Suhang Wang, Dongwon Lee, and Huan Liu

  31. [39]

    Zirui Song, Yaohang Li, Meng Fang, Zhenhao Chen, Zecheng Shi, and Yuan Huang. 2024. Mmac-copilot: Multi-modal agent collaboration operating system copilot.arXiv e-prints(2024), arXiv–2404

  32. [40]

    Kate Starbird, Jim Maddock, Mania Orand, Peg Achterman, and Robert M Mason

  33. [41]

    Yashar Talebirad and Amirhossein Nadiri. 2023. Multi-agent collaboration: Har- nessing the power of intelligent llm agents.arXiv preprint arXiv:2306.03314 (2023)

  34. [42]

    Petter Törnberg, Diliara Valeeva, Justus Uitermark, and Christopher Bail. 2023. Simulating social media using large language models to evaluate alternative news feed algorithms.arXiv preprint arXiv:2310.05984(2023)

  35. [43]

    Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph Attention Networks. InInternational Con- ference on Learning Representations

  36. [44]

    Yaqing Wang, Fenglong Ma, Zhiwei Jin, Ye Yuan, Guangxu Xun, Kishlay Jha, Lu Su, and Jing Gao. 2018. Eann: Event adversarial neural networks for multi-modal fake news detection. InProceedings of the 24th acm sigkdd international conference on knowledge discovery & data mining. 849–857

  37. [45]

    Biao Wu, Yanda Li, Meng Fang, Zirui Song, Zhiwei Zhang, Yunchao Wei, and Ling Chen. 2024. Foundations and recent trends in multimodal mobile agents: A survey.arXiv preprint arXiv:2411.02006(2024)

  38. [46]

    Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Shaokun Zhang, Erkang Zhu, Beibin Li, Li Jiang, Xiaoyun Zhang, and Chi Wang. 2023. Autogen: Enabling next-gen llm applications via multi-agent conversation framework.arXiv preprint arXiv:2308.08155(2023)

  39. [47]

    Canwen Xu, Yichong Xu, Shuohang Wang, Yang Liu, Chenguang Zhu, and Julian McAuley. 2023. Small models are valuable plug-ins for large language models. arXiv preprint arXiv:2305.08848(2023). The Truth Becomes Clearer Through Debate!Multi-Agent Systems with Large Language Models...

  40. [48]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, et al . 2024. Qwen2. 5 Technical Report.arXiv preprint arXiv:2412.15115(2024)

  41. [49]

    Feng Yu, Qiang Liu, Shu Wu, Liang Wang, Tieniu Tan, et al. 2017. A Convolutional Approach for Misinformation Identification.. InIJCAI. 3901–3907

  42. [50]

    Liu Yuhan, Chen Xiuying, and Yan Rui. 2023. Unleashing the power of large models: Exploring human-machine conversations. InProceedings of the 22nd Chinese National Conference on Computational Linguistics (Volume 2: Frontier Forum). 16–29

  43. [51]

    Juntian Zhang, Chuanqi Cheng, Yuhan Liu, Wei Liu, Jian Luan, and Rui Yan. 2025. Weaving Context Across Images: Improving Vision-Language Models through Focus-Centric Visual Chains.arXiv preprint arXiv:2504.20199(2025)

  44. [52]

    Jintian Zhang, Xin Xu, Ningyu Zhang, Ruibo Liu, Bryan Hooi, and Shumin Deng

  45. [53]

    Xueyao Zhang, Juan Cao, Xirong Li, Qiang Sheng, Lei Zhong, and Kai Shu. 2021. Mining dual emotion for fake news detection. InProceedings of the web conference

  46. [54]

    Xiaoqing Zhang, Xiuying Chen, Yuhan Liu, Jianzhou Wang, Zhenxing Hu, and Rui Yan. 2024. A large-scale time-aware agents simulation for influencer selection in digital advertising campaigns.arXiv preprint arXiv:2411.01143(2024)

  47. [55]

    Xiaoqing Zhang, Xiuying Chen, Yuhan Liu, Jianzhou Wang, Zhenxing Hu, and Rui Yan. 2024. SAGraph: A Large-scale Text-Rich Social Graph Dataset for Advertising Campaigns.arXiv preprint arXiv:2403.15105(2024)

  48. [56]

    Xiaoqing Zhang, Yuhan Liu, Flood Sung, Xiuying Chen, and Rui Yan. 2024. Think- ing Before Running! Efficient Code Generation with Thorough Exploration and Optimal Refinement.arXiv preprint arXiv:2502.17442(2024)

  49. [57]

    Zheng Zhang, Qi Liu, Zirui Hu, Yi Zhan, Zhenya Huang, Weibo Gao, and Qingyang Mao. 2024. Enhancing fairness in meta-learned user modeling via adaptive sampling. InProceedings of the ACM Web Conference 2024. 3241–3252

  50. [58]

    InICLR 2024 Workshop on Large Language Model (LLM) Agents

    Exploring Collaboration Mechanisms for LLM Agents: A Social Psychology View. InICLR 2024 Workshop on Large Language Model (LLM) Agents

  51. [59]

    Zheng Zhang, Wei Song, Qi Liu, Qingyang Mao, Yiyan Wang, Weibo Gao, Zhenya Huang, Shijin Wang, and Enhong Chen. [n. d.]. Towards Accurate and Fair Cognitive Diagnosis via Monotonic Data Augmentation. InThe Thirty-eighth Annual Conference on Neural Information Processing Systems

  52. [60]

    Jiaqi Zheng, Xi Zhang, Sanchuan Guo, Quan Wang, Wenyu Zang, and Yongdong Zhang. 2022. MFAN: Multi-modal Feature-enhanced Attention Networks for Rumor Detection. IJCAI

  53. [61]

    Yongchun Zhu, Qiang Sheng, Juan Cao, Shuokai Li, Danding Wang, and Fuzhen Zhuang. 2022. Generalizing to the future: Mitigating entity bias in fake news detection. InProceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval....

  54. [64]

    Zheng Zhang, Qi Liu, Hao Jiang, Fei Wang, Yan Zhuang, Le Wu, Weibo Gao, and Enhong Chen. 2023. FairLISA: fair user modeling with limited sensitive attributes information. InProceedings of the 37th International Conference on Neural Information Processing Systems. 41432–41450

  55. [2013]

    InProceedings of the 22nd international conference on World Wide Web

    Faking sandy: characterizing and identifying fake images on twitter during hurricane sandy. InProceedings of the 22nd international conference on World Wide Web. 729–736

  56. [2014]

    Rumors, false flags, and digital vigilantes: Misinformation on twitter after the 2013 boston marathon bombing.IConference 2014 proceedings(2014)

  57. [2020]

    Fakenewsnet: A data repository with news content, social context, and spatiotemporal information for studying fake news on social media.Big data8, 3 (2020), 171–188

  58. [2023]

    Improving factuality and reasoning in language models through multiagent debate.arXiv preprint arXiv:2305.14325(2023)

  59. [2024]

    InProceedings of the Thirty-Third International Joint Conference on Artificial Intelligence

    From skepticism to acceptance: simulating the attitude dynamics toward fake news. InProceedings of the Thirty-Third International Joint Conference on Artificial Intelligence. 7886–7894

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.