Genomic Convergence Verification: A Multi-Agent Framework for Hallucination Detection in Retrieval-Augmented Generation. Overcoming Gödel's Limitation Through Multi-System Convergence.

Владислав Витальевич Тишков

ПРЕПРИНТ

Эта статья является препринтом и не была отрецензирована.
О результатах, изложенных в препринтах, не следует сообщать в СМИ как о проверенной информации.

Genomic Convergence Verification: A Multi-Agent Framework for Hallucination Detection in Retrieval-Augmented Generation. Overcoming Gödel's Limitation Through Multi-System Convergence.

В. В. Тишков

2026-08-23

https://doi.org/10.24108/preprints-3116223

We present Genomic Convergence Verification (GCV), a novel theoretical framework and practical implementation for detecting hallucinations in Retrieval-Augmented Generation (RAG) systems. The framework addresses a fundamental limitation: any single verification system is bounded by Gödel's incompleteness theorems, meaning there will always exist statements whose truth value cannot be determined within that system. Our approach deploys multiple independent verification agents and identifies their absolute convergence points—termed the genomic code of truth—where all agents unanimously agree with high confidence. We prove that as the number of independent verification systems increases, the probability of a statement being undecidable across all systems asymptotically approaches zero. The implementation comprises three specialized verification agents (fact checker, contradiction detector, completeness evaluator), a differential convergence equation for modeling confidence evolution, an apogee detection algorithm for finding optimal convergence points, and a genomic code extraction mechanism for identifying absolute truth patterns. The framework is modular, extensible, and available as open-source software. It represents a step toward trustworthy AI systems where verification is not bounded by the mathematical limitations of any single approach.

Ссылка для цитирования:

Тишков В. В. 2026. Genomic Convergence Verification: A Multi-Agent Framework for Hallucination Detection in Retrieval-Augmented Generation. Overcoming Gödel's Limitation Through Multi-System Convergence. PREPRINTS.RU. https://doi.org/10.24108/preprints-3116223

Список литературы

1. K. Gödel. Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I. Monatshefte für Mathematik und Physik, 38:173-198, 1931.

2. P. Lewis, E. Perez, A. Piktus, F. Petroni, V. Karpukhin, N. Goyal, H. Küttler, M. Lewis, W. Yih, T. Rocktäschel, S. Riedel, and D. Kiela. Retrieval-augmented generation for knowledge-intensive NLP tasks. In Advances in Neural Information Processing Systems (NeurIPS), volume 33, pages 9459-9474, 2020.

3. J. Maynez, S. Narayan, B. Bohnet, and R. McDonald. On faithfulness and factuality in abstractive summarization. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL), pages 1906-1919, 2020.

4. K. Shuster, S. Poff, M. Chen, D. Kiela, and J. Weston. Retrieval augmentation reduces hallucination in conversation. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 3784-3803, 2021.

5. W. Kryściński, B. McCann, C. Xiong, and R. Socher. Evaluating the factual consistency of abstractive text summarization. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 9332-9346, 2020.

6. P. Laban, T. Schnabel, P. N. Bennett, and M. A. Hearst. SummaC: Re-visiting NLI-based models for inconsistency detection in summarization. Transactions of the Association for Computational Linguistics, 10:163-177, 2022.

7. Y. Zha, Y. Yang, R. Li, and Z. Hu. AlignScore: Evaluating factual consistency with a unified alignment function. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL), 2023.

8. P. Manakul, A. Liusie, and M. J. F. Gales. SelfCheckGPT: Zero-resource black-box hallucination detection for generative large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023.

9. R. Cohen, M. Hamri, M. Geva, and A. Globerson. LM vs LM: Detecting factual errors via cross examination. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023.

10. Y. Du, S. Li, A. Torralba, J. Tenenbaum, and I. Mordatch. Improving factuality and reasoning in language models through multiagent debate. arXiv preprint arXiv:2305.14325, 2023.

11. J. Thorne, A. Vlachos, C. Christodoulopoulos, and A. Mittal. FEVER: a large-scale dataset for fact extraction and verification. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT), pages 809-819, 2018.

12. Z. Guo, M. Schlichtkrull, and A. Vlachos. A survey on automated fact-checking. Transactions of the Association for Computational Linguistics, 10:178-206, 2022.

13. N. Reimers and I. Gurevych. Sentence-BERT: Sentence embeddings using Siamese BERT-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing (EMNLP-IJCNLP), pages 3982-3992, 2019.

14. 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 Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2023.

15. Y. Liu, M. Ott, N. Goyal, J. Du, M. Joshi, D. Chen, O. Levy, M. Lewis, L. Zettlemoyer, and V. Stoyanov. RoBERTa: A robustly optimized BERT pretraining approach. arXiv preprint arXiv:1907.11692, 2019.

ПРЕПРИНТ

ИНФОРМАЦИЯ

Использование куки-файлов