Multi-Agent Cross-Examination for Hallucination Detection and Evidence Localization in Large Language Models
Keywords:
Hallucination Detection, Evidence Localization, Multi-Agent Systems, Cross-Examination, Self-Consistency, Halueval, CalibrationAbstract
Response-level hallucination labels do not identify the decisive text, while repeated judgments add cost without a distinct adversarial role. We evaluate a Detector–Critic–Arbiter framework on 4,507 HaluEval General records. The Detector estimates response risk from lexical and structural views; the Critic scores 27,335 evidence units; and an out-of-fold Arbiter combines probabilities, disagreement, evidence concentration, and structure. A stratified 70/15/15 split, development-tuned thresholds, 300 bootstrap replicates, and paired held-out tests support evaluation. Five-run self-consistency achieved the best macro-F1 (0.642), AUROC (0.721), Brier score (0.211), and accuracy at 80% selective coverage (0.793). Cross-examination achieved the highest AUPRC (0.365), evidence-unit precision (0.667), character precision (0.660), and Hit@1 (0.739), with macro-F1 0.625, ECE 0.290, and character F1 0.444 versus 0.451 for self-consistency. Ablations showed that the learned Arbiter improved ranking over a simple mean, but explicit disagreement features did not improve AUPRC. Capability and refusal errors were easier than semantic or factual errors. Thus, self-consistency is stronger for balanced screening, whereas cross-examination yields more selective, better-ranked evidence for review.
References
[1] Z. Ji et al., “Survey of hallucination in natural language generation,” ACM Comput. Surv., vol. 55, no. 12, Art. no. 248, pp. 1–38, 2023.
[2] J. Nie and D. Zheng, “Ambiguity-aware HDFS log anomaly detection with retrieval-augmented failure narratives and selective refusal,” J. Adv. Comput. Syst., vol. 3, no. 1, pp. 66–80, Jan. 2023, doi: 10.69987/JACS.2023.30105.
[3] J. Li, X. Cheng, W. X. Zhao, J.-Y. Nie, and J.-R. Wen, “HaluEval: A large-scale hallucination evaluation benchmark for large language models,” in Proc. 2023 Conf. Empirical Methods in Natural Language Processing, 2023, pp. 6449–6464.
[4] Z. S. Zhong, R. Ma, and H. Zhao, “Human-uncertainty distillation for calibrated vision models on CIFAR-10H,” J. Adv. Comput. Syst., vol. 3, no. 2, pp. 77–89, Feb. 2023, doi: 10.69987/JACS.2023.30206.
[5] S. Lin, J. Hilton, and O. Evans, “TruthfulQA: Measuring how models mimic human falsehoods,” in Proc. 60th Annu. Meeting Assoc. Comput. Linguistics, 2022, pp. 3214–3252.
[6] P. Manakul, A. Liusie, and M. J. F. Gales, “SelfCheckGPT: Zero-resource black-box hallucination detection for generative large language models,” in Proc. 2023 Conf. Empirical Methods in Natural Language Processing, 2023, pp. 9004–9017.
[7] S. Zhou, Z. Li, and E. Wang, “Evidence-grounded RAG for tokenized trade receivable disclosure QA under U.S. capital market standards,” J. Adv. Comput. Syst., vol. 3, no. 7, pp. 41–57, Jul. 2023, doi: 10.69987/JACS.2023.30704.
[8] S. Min et al., “FActScore: Fine-grained atomic evaluation of factual precision in long form text generation,” in Proc. 2023 Conf. Empirical Methods in Natural Language Processing, 2023, pp. 12076–12100.
[9] S. Zhou, Z. Li, and E. Wang, “Long-document RAG for contractual and insurance clause analysis in receivables RWA structures,” J. Adv. Comput. Syst., vol. 4, no. 8, pp. 88–104, Aug. 2024, doi: 10.69987/JACS.2024.40810.
[10] A. R. Fabbri, C.-S. Wu, W. Liu, and C. Xiong, “QAFactEval: Improved QA-based factual consistency evaluation for summarization,” in Proc. 2022 Conf. North American Chapter Assoc. Comput. Linguistics, 2022, pp. 2587–2601.
[11] A. Wang, K. Cho, and M. Lewis, “Asking and answering questions to evaluate the factual consistency of summaries,” in Proc. 58th Annu. Meeting Assoc. Comput. Linguistics, 2020, pp. 5008–5020.
[12] Y. Li, “Findable then explainable: Retrieval–summary integration for code intelligence on a lightweight CodeSearchNet subset,” J. Adv. Comput. Syst., vol. 4, no. 7, pp. 65–82, Jul. 2024, doi: 10.69987/JACS.2024.40706.
[13] P. Laban, T. Schnabel, P. N. Bennett, and M. A. Hearst, “SummaC: Re-visiting NLI-based models for inconsistency detection in summarization,” Trans. Assoc. Comput. Linguistics, vol. 10, pp. 163–177, 2022.
[14] W. Kryściński, B. McCann, C. Xiong, and R. Socher, “Evaluating the factual consistency of abstractive text summarization,” in Proc. 2020 Conf. Empirical Methods in Natural Language Processing, 2020, pp. 9332–9346.
[15] J. Thorne, A. Vlachos, C. Christodoulopoulos, and A. Mittal, “FEVER: A large-scale dataset for fact extraction and verification,” in Proc. 2018 Conf. North American Chapter Assoc. Comput. Linguistics, 2018, pp. 809–819.
[16] J. Maynez, S. Narayan, B. Bohnet, and R. McDonald, “On faithfulness and factuality in abstractive summarization,” in Proc. 58th Annu. Meeting Assoc. Comput. Linguistics, 2020, pp. 1906–1919.
[17] J. DeYoung et al., “ERASER: A benchmark to evaluate rationalized NLP models,” in Proc. 58th Annu. Meeting Assoc. Comput. Linguistics, 2020, pp. 4443–4458.
[18] Z. Li, S. Zhou, and Z. Zhou, “Financial risk dashboard design for institutional RWA investors: Visual hierarchy, chart comprehension, and explainability in FinChart-Bench,” Int. J. Graph. Des., vol. 3, no. 1, pp. 196–210, May 2025, doi: 10.51903/ijgd.v3i1.3715.
[19] T. Lei, R. Barzilay, and T. Jaakkola, “Rationalizing neural predictions,” in Proc. 2016 Conf. Empirical Methods in Natural Language Processing, 2016, pp. 107–117.
[20] H. Wang, Y. Ren, and X. Chang, “Layout-aware progressive PDF rendering: AI prioritization of PDF slices to reduce time-to-functional-first-frame on FUNSD,” J. Technol. Informatics Eng., vol. 4, no. 2, pp. 425–446, Aug. 2025, doi: 10.51903/jtie.v4i2.523.
[21] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. 2019 Conf. North American Chapter Assoc. Comput. Linguistics, 2019, pp. 4171–4186.
[22] R. Ma, L. Zhang, and T. Song, “Computer-vision-informed visual explanation cards for autonomous-driving traffic-sign alerts: Localization, classification, and retrieved evidence on GTSDB,” Int. J. Graph. Des., vol. 3, no. 2, pp. 455–473, Oct. 2025, doi: 10.51903/ijgd.v3i2.3991.
[23] S. R. Bowman, G. Angeli, C. Potts, and C. D. Manning, “A large annotated corpus for learning natural language inference,” in Proc. 2015 Conf. Empirical Methods in Natural Language Processing, 2015, pp. 632–642.
[24] X. Wang et al., “Self-consistency improves chain of thought reasoning in language models,” in Proc. Int. Conf. Learning Representations, 2023.
[25] Z. S. Zhong and S. Ling, “Improved theoretical guarantee for rank aggregation via spectral method,” Inf. Inference, vol. 13, no. 3, Art. no. iaae020, Sep. 2024, doi: 10.1093/imaiai/iaae020.
[26] Y. Du, S. Li, A. Torralba, J. B. Tenenbaum, and I. Mordatch, “Improving factuality and reasoning in language models through multiagent debate,” arXiv:2305.14325, 2023.
[27] C.-M. Chan et al., “ChatEval: Towards better LLM-based evaluators through multi-agent debate,” arXiv:2308.07201, 2023.
[28] Y. He, Y. Pan, Y. Wang, S. Du, and Q. Xin, “Intelligent fault analysis with AIOps technology,” J. Theory Pract. Eng. Sci., vol. 4, no. 1, pp. 94–100, Feb. 2024, doi: 10.53469/jtpes.2024.04(01).13.
[29] A. Madaan et al., “Self-Refine: Iterative refinement with self-feedback,” in Adv. Neural Inf. Process. Syst., vol. 36, 2023, pp. 46534–46594.
[30] N. Shinn, F. Cassano, A. Gopinath, K. Narasimhan, and S. Yao, “Reflexion: Language agents with verbal reinforcement learning,” in Adv. Neural Inf. Process. Syst., vol. 36, 2023, pp. 8634–8652.
[31] H. Zhang, “LLM-driven CI failure diagnosis and automated repair: From GitHub Actions logs to patch recommendation,” J. Technol. Informatics Eng., vol. 4, no. 1, pp. 190–214, Feb. 2025.
[32] Z. Gou et al., “CRITIC: Large language models can self-correct with tool-interactive critiquing,” arXiv:2305.11738, 2023.
[33] S. Dhuliawala et al., “Chain-of-verification reduces hallucination in large language models,” arXiv:2309.11495, 2023.
[34] Q. Xin, “Explaining OpenStack failure-injection log anomalies with retrieved normal prototypes,” Emerg. Inf. Sci. Technol., vol. 6, no. 2, pp. 125–146, Nov. 2025, doi: 10.18196/eist.v6i2.31232.
[35] Y.-S. Chuang et al., “DoLa: Decoding by contrasting layers improves factuality in large language models,” arXiv:2309.03883, 2023.
[36] I.-C. Chern et al., “FacTool: Factuality detection in generative AI—A tool augmented framework for multi-task and multi-domain scenarios,” arXiv:2307.13528, 2023.
[37] C. D. Manning, P. Raghavan, and H. Schütze, Introduction to Information Retrieval. Cambridge, U.K.: Cambridge Univ. Press, 2008.
[38] Q. Xin, Z. Xu, L. Guo, F. Zhao, and B. Wu, “IoT traffic classification and anomaly detection method based on deep autoencoders,” in Proc. 6th Int. Conf. Computing and Data Science, 2024, pp. 64–70, doi: 10.54254/2755-2721/69/20241511.
[39] B. Wang, Y. He, Z. Shui, Q. Xin, and H. Lei, “Predictive optimization of DDoS attack mitigation in distributed systems using machine learning,” in Proc. 6th Int. Conf. Computing and Data Science, 2024, pp. 94–99, doi: 10.54254/2755-2721/64/20241350.
[40] F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J. Mach. Learn. Res., vol. 12, no. 85, pp. 2825–2830, 2011.
[41] Y. Wang, S. Du, Q. Xin, Y. He, and W. Qian, “Autonomous driving system driven by artificial intelligence perception fusion,” Acad. J. Sci. Technol., vol. 9, no. 2, pp. 193–198, Feb. 2024, doi: 10.54097/e0b9ak47.
[42] J. Nie and D. Zheng, “Noisy-neighbor-aware VM degradation risk modeling with unsupervised residual fusion,” J. Adv. Comput. Syst., vol. 4, no. 4, pp. 112–123, Apr. 2024, doi: 10.69987/JACS.2024.40409.
[43] Q. Xin, “Uncertainty-aware late fusion for 3D perception: Confidence calibration and fusion-rule learning,” J. Technol. Informatics Eng., vol. 4, no. 1, pp. 215–238, Apr. 2025, doi: 10.51903/jtie.v4i1.485.
[44] Z. S. Zhong and S. Ling, “Uncertainty quantification of spectral estimator and MLE for orthogonal group synchronization,” arXiv:2408.05944, Aug. 2024.
[45] G. W. Brier, “Verification of forecasts expressed in terms of probability,” Monthly Weather Rev., vol. 78, no. 1, pp. 1–3, 1950.
[46] C. Guo, G. Pleiss, Y. Sun, and K. Q. Weinberger, “On calibration of modern neural networks,” in Proc. 34th Int. Conf. Machine Learning, 2017, pp. 1321–1330.
[47] B. Efron and R. J. Tibshirani, An Introduction to the Bootstrap. New York, NY, USA: Chapman & Hall, 1993.
[48] Q. McNemar, “Note on the sampling error of the difference between correlated proportions or percentages,” Psychometrika, vol. 12, no. 2, pp. 153–157, 1947.
[49] K. Xu, H. Zhou, H. Zheng, M. Zhu, and Q. Xin, “Intelligent classification and personalized recommendation of e-commerce products based on machine learning,” in Proc. 6th Int. Conf. Computing and Data Science, 2024, pp. 143–149, doi: 10.54254/2755-2721/64/20241365.
[50] H. Zhang, “Counterfactual learning-to-rank for ads: Off-policy evaluation on the Open Bandit Dataset,” J. Adv. Comput. Syst., vol. 5, no. 12, pp. 1–11, 2025.
[51] J. Wang, Q. Xin, Y. Liu, J. Wang, and T. Yang, “Predicting enterprise marketing decision making with intelligent data-driven approaches,” J. Ind. Eng. Appl. Sci., vol. 2, no. 3, pp. 12–19, Jun. 2024, doi: 10.5281/zenodo.11357252.
[52] Z. Ling, Q. Xin, Y. Lin, G. Su, and Z. Shui, “Optimization of autonomous driving image detection based on RFAConv and triplet attention,” in Proc. 2nd Int. Conf. Software Engineering and Machine Learning, 2024, pp. 210–217, doi: 10.54254/2755-2721/77/2024MA0067.
[53] Q. Xin, R. Song, Z. Wang, Z. Xu, and F. Zhao, “Enhancing bank credit risk management using the C5.0 decision tree algorithm,” J. Comput. Technol. Appl. Math., vol. 1, no. 4, pp. 100–107, Nov. 2024, doi: 10.5281/zenodo.14032041.
[54] T. Yang, Q. Xin, X. Zhan, S. Zhuang, and H. Li, “Enhancing financial services through big data and AI-driven customer insights and risk analysis,” J. Knowl. Learn. Sci. Technol., vol. 3, no. 3, pp. 53–62, Jul. 2024, doi: 10.60087/jklst.vol3.n3.p53-62.
[55] H. Zhang, “Risk-aware budget-constrained auto-bidding under first-price RTB: A distributional constrained deep reinforcement learning framework,” J. Adv. Comput. Syst., vol. 4, no. 6, pp. 30–47, Jun. 2024, doi: 10.69987/JACS.2024.40603.
[56] H. Zhou and K. Zhang, “News-based uncertainty and macro-market fusion for VIX direction forecasting: Evidence from a 2015–2024 FRED panel,” J. Technol. Informatics Eng., vol. 4, no. 2, pp. 487–501, Aug. 2025, doi: 10.51903/jtie.v4i2.540.
[57] Y. Lu, H. Zhou, and Y. Zhang, “A constrained, data-driven budgeting framework integrating macro demand forecasting and marketing response modeling,” J. Technol. Informatics Eng., vol. 4, no. 3, pp. 493–520, Dec. 2025, doi: 10.51903/jtie.v4i3.466.
[58] Z. Zhong, M. Zheng, H. Mai, J. Zhao, and X. Liu, “Cancer image classification based on DenseNet model,” J. Phys.: Conf. Ser., vol. 1651, no. 1, Art. no. 012143, 2020, doi: 10.1088/1742-6596/1651/1/012143.
[59] J. Chen, J. Xiong, Y. Wang, Q. Xin, and H. Zhou, “Implementation of an AI-based MRD evaluation and prediction model for multiple myeloma,” Front. Comput. Intell. Syst., vol. 6, no. 3, pp. 127–131, 2024, doi: 10.54097/zJ4MnbWW.
[60] Q. Xin, “Hybrid cloud architecture for efficient and cost-effective large language model deployment,” J. Inf. Syst. Informatics, vol. 7, no. 3, pp. 2182–2195, Sep. 2025, doi: 10.51519/journalisi.v7i3.1170.
[61] B. Zhou, H. Wang, and X. Chang, “Distilling VMAF into an edge-deployable quality predictor: A pilot shot-level proxy with LLM-ready quality tokens,” J. Technol. Informatics Eng., vol. 4, no. 2, pp. 447–463, Aug. 2025, doi: 10.51903/jtie.v4i2.522.
[62] A. G. S. Raj et al., “Impact of bilingual CS education on student learning and engagement in a data structures course,” in Proc. 19th Koli Calling Int. Conf. Comput. Educ. Res., 2019, pp. 1–10, doi: 10.1145/3364510.3364518.
[63] Z. S. Zhong, X. Pan, and Q. Lei, “Bridging domains with approximately shared features,” in Proc. 28th Int. Conf. Artificial Intelligence and Statistics, PMLR, vol. 258, 2025, pp. 559–567.
[64] L. Zhang, R. Ma, and P. Greg, “Digital-twin dispatching for urban mobility via spatio-temporal transformers and offline reinforcement learning,” J. Technol. Informatics Eng., vol. 4, no. 2, pp. 337–363, Aug. 2025, doi: 10.51903/jtie.v4i2.501.
[65] H. Zhang, “DriftGuard: Multi-signal drift early warning and safe re-training/rollback for CTR/CVR models,” J. Adv. Comput. Syst., vol. 3, no. 7, pp. 24–40, Jul. 2023, doi: 10.69987/JACS.2023.30703
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Xu Zhou, Michelle Carter, Kai Wang, Andrew Mitchell (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
Creative Commons Attribution 4.0 International (CC BY 4.0).
This work is licensed under a 