<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.3 20210610//EN" "JATS-journalpublishing1-3.dtd">
<article article-type="research-article" dtd-version="1.3" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xml:lang="ru"><front><journal-meta><journal-id journal-id-type="publisher-id">ntv</journal-id><journal-title-group><journal-title xml:lang="ru">Научно-технический вестник информационных технологий, механики и оптики</journal-title><trans-title-group xml:lang="en"><trans-title>Scientific and Technical Journal of Information Technologies, Mechanics and Optics</trans-title></trans-title-group></journal-title-group><issn pub-type="ppub">2226-1494</issn><issn pub-type="epub">2500-0373</issn><publisher><publisher-name>Университет ИТМО</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="doi">10.17586/2226-1494-2026-26-1-69-76</article-id><article-id custom-type="elpub" pub-id-type="custom">ntv-565</article-id><article-categories><subj-group subj-group-type="heading"><subject>Research Article</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="ru"><subject>КОМПЬЮТЕРНЫЕ СИСТЕМЫ И ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="en"><subject>COMPUTER SCIENCE</subject></subj-group></article-categories><title-group><article-title>Подход к применению больших языковых моделей с дополненным поиском для повышения интерпретируемости моделей машинного обучения</article-title><trans-title-group xml:lang="en"><trans-title>An approach to using large language models with augmented search to improve the interpretability of machine learning models</trans-title></trans-title-group></title-group><contrib-group><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0009-0004-1664-1512</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Иванов</surname><given-names>Р. Д.</given-names></name><name name-style="western" xml:lang="en"><surname>Ivanov</surname><given-names>R. D.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Иванов Роман Дмитриевич — старший помощник начальникаотдела, младший научный сотрудник</p><p>Санкт-Петербург, 197198</p></bio><bio xml:lang="en"><p>Roman D. Ivanov — Senior Assistant to the Head of the Department, Junior Research</p><p>Saint Petersburg, 197198</p></bio><email xlink:type="simple">vka@mil.ru</email><xref ref-type="aff" rid="aff-1"/></contrib><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0000-0002-9955-2694</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Менисов</surname><given-names>А. Б.</given-names></name><name name-style="western" xml:lang="en"><surname>Menisov</surname><given-names>A. B.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Менисов Артем Бакытжанович — доктор технических наук,старший преподаватель</p><p>Санкт-Петербург, 197198</p><p>sc 57220815185</p></bio><bio xml:lang="en"><p>Artem B. Menisov — D.Sc., Senior Lecturer</p><p>Saint Petersburg, 197198</p><p>sc 57220815185</p></bio><email xlink:type="simple">men.arty@yandex.ru</email><xref ref-type="aff" rid="aff-1"/></contrib><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0009-0004-4247-5810</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Мишуков</surname><given-names>О. А.</given-names></name><name name-style="western" xml:lang="en"><surname>Mishukov</surname><given-names>O. A.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Мишуков Олег Александрович — кандидат технических наук,старший преподаватель</p><p>Санкт-Петербург, 197198</p></bio><bio xml:lang="en"><p>Oleg A. Mishukov — PhD, Senior Lecturer</p><p>Saint Petersburg, 197198</p></bio><email xlink:type="simple">vka@mil.ru</email><xref ref-type="aff" rid="aff-1"/></contrib><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0000-0002-6807-2954</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Сабиров</surname><given-names>Т. Р.</given-names></name><name name-style="western" xml:lang="en"><surname>Sabirov</surname><given-names>T. R.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Сабиров Тимур Римович — кандидат технических наук, старший преподаватель</p><p>Санкт-Петербург, 197198</p><p>sc 57188236500</p></bio><bio xml:lang="en"><p>Timur R. Sabirov — PhD, Senior Lecturer</p><p>Saint Petersburg, 197198</p><p>sc 57188236500</p></bio><email xlink:type="simple">vka@mil.ru</email><xref ref-type="aff" rid="aff-1"/></contrib></contrib-group><aff-alternatives id="aff-1"><aff xml:lang="ru"><institution>Военно-космическая академия имени А.Ф. Можайского</institution><country>Россия</country></aff><aff xml:lang="en"><institution>Mozhaisky Military Aerospace Academy</institution><country>Russian Federation</country></aff></aff-alternatives><pub-date pub-type="collection"><year>2026</year></pub-date><pub-date pub-type="epub"><day>25</day><month>02</month><year>2026</year></pub-date><volume>26</volume><issue>1</issue><fpage>69</fpage><lpage>76</lpage><permissions><copyright-statement>Copyright &amp;#x00A9; Иванов Р.Д., Менисов А.Б., Мишуков О.А., Сабиров Т.Р., 2026</copyright-statement><copyright-year>2026</copyright-year><copyright-holder xml:lang="ru">Иванов Р.Д., Менисов А.Б., Мишуков О.А., Сабиров Т.Р.</copyright-holder><copyright-holder xml:lang="en">Ivanov R.D., Menisov A.B., Mishukov O.A., Sabirov T.R.</copyright-holder><license xml:lang="ru" license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/" xlink:type="simple"><license-p>Данная работа распространяется под лицензией Creative Commons Attribution 4.0.</license-p></license><license xml:lang="en" license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/" xlink:type="simple"><license-p>This work is licensed under a Creative Commons Attribution 4.0 License.</license-p></license></permissions><self-uri xlink:href="https://ntv.elpub.ru/jour/article/view/565">https://ntv.elpub.ru/jour/article/view/565</self-uri><abstract><p>Введение. Интерпретируемость моделей машинного обучения является ключевым условием доверия при внедрении технологий искусственного интеллекта. Традиционные методы объяснения обеспечивают формальную интерпретацию и представляют фрагментарные и трудно интерпретируемые результаты, что снижает прозрачность принятия решений. Для решения этой проблемы предлагается подход, основанный на использовании больших языковых моделей в сочетании с технологией дополненного поиска, обеспечивающей привлечение внешних знаний и повышающей связность объяснений. Отличительной особенностью подхода является ориентация на семантическую согласованность, стабильность интерпретаций и понятность для пользователя. Метод. В представленном подходе большая языковая модель выступает в качестве интерпретатора, формирующего объяснения на основе исходных данных модели и внешних знаний, получаемых через дополненный поиск. Подход включает механизм генерации человеческо-читаемых объяснений, верификацию интерпретируемых признаков (токенов) и оценку устойчивости выводов. В результате формируются выводы, согласованные с контекстом предметной области. Основные результаты. Эффективность подхода проверена на данных MITRE ATT&amp;CK, представляющих стандартизированную информацию о киберугрозах. Сравнение с методом SHapley Additive exPlanations показало, что предложенный подход обеспечивает более высокую семантическую согласованность объяснений и большую устойчивость оценки важности признаков при сохранении сопоставимого уровня точности интерпретации. Экспериментальные результаты показали преимущество интерпретатора на основе больших языковых моделей в объяснимости и восприятии человеком.Обсуждение. Разработанный подход в сравнении с традиционными методами обеспечивает понятные и контекстно обоснованные объяснения. Полученные результаты делают предложенный метод перспективным для применения в системах, где критично объяснение причин принятия решений. Дальнейшие направления исследований включают интеграцию подхода в реальные системы принятия решений, исследование автоматического контроля достоверности объяснений и адаптацию подхода к различным архитектурам больших языковых моделей.</p></abstract><trans-abstract xml:lang="en"><p>Interpretability of machine learning models is a key requirement for robustness in the implementation of artificial intelligence technologies. Traditional explanation methods provide formal interpretation and produce results that are fragmented and hard to interpret, reducing the transparency of decision making. To address this problem, a modern approach is proposed based on the use of large language models combined with augmented search, which ensures the involvement of external knowledge and increases the coherence of sequences. A distinctive feature of this approach is the focus on semantic consistency, the stability of interpretations, and user comprehensibility. In the presented approach, a large language model acts as an interpreter, generating explanations based on the model input data and external knowledge obtained through augmented search. The approach includes a mechanism for generating human-readable observations, verifying interpretable features (tokens), and assessing the robustness of inferences. As a result, inferences are generated that are consistent with the context of the subject domain. The effectiveness of the proposed approach is tested using MITRE ATT&amp;CK data, which provides standardized information on cyber threats. A comparison with the SHapley Additive exPlanations method showed that the proposed approach provides higher semantic consistency of consequences and greater robustness of feature importance assessment while maintaining advanced level accuracy. Experimental results obtained by the interpreter based on large language models are in terms of human comprehensibility and perception. The developed approach, when considered with conservative methods, provides understandable and context-based explanations. The obtained results make the proposed method promising for application in economics where critical explanations of decision-making are essential. Future research includes integrating the solutions into real systems, investigating automatic validation of results, and adapting them to various large language model architectures.</p></trans-abstract><kwd-group xml:lang="ru"><kwd>большая языковая модель</kwd><kwd>интерпретируемость</kwd><kwd>дополненный поиск</kwd><kwd>семантическая согласованность</kwd></kwd-group><kwd-group xml:lang="en"><kwd>large language model</kwd><kwd>interpretability</kwd><kwd>augmented search</kwd><kwd>semantic consistency</kwd></kwd-group></article-meta></front><back><ref-list><title>References</title><ref id="cit1"><label>1</label><citation-alternatives><mixed-citation xml:lang="ru">Yang Y., Zhang Y., Sun D., He W., Wei Y. Navigating the landscape of AI literacy education: insights from a decade of research (2014– 2024) // Humanities and Social Sciences Communications. 2025. V. 12. N 1. P. 374. https://doi.org/10.1057/s41599-025-04583-8</mixed-citation><mixed-citation xml:lang="en">Yang Y., Zhang Y., Sun D., He W., Wei Y. Navigating the landscape of AI literacy education: insights from a decade of research (2014– 2024) // Humanities and Social Sciences Communications. 2025. V. 12. N 1. P. 374. https://doi.org/10.1057/s41599-025-04583-8</mixed-citation></citation-alternatives></ref><ref id="cit2"><label>2</label><citation-alternatives><mixed-citation xml:lang="ru">Mon-Williams R., Li G., Long R., Du W., Lucas C.G. Embodied large language models enable robots to complete complex tasks in unpredictable environments // Nature Machine Intelligence. 2025. V. 7. N 4. P. 592–601. https://doi.org/10.1038/s42256-025-01005-x</mixed-citation><mixed-citation xml:lang="en">Mon-Williams R., Li G., Long R., Du W., Lucas C.G. Embodied large language models enable robots to complete complex tasks in unpredictable environments // Nature Machine Intelligence. 2025. V. 7. N 4. P. 592–601. https://doi.org/10.1038/s42256-025-01005-x</mixed-citation></citation-alternatives></ref><ref id="cit3"><label>3</label><citation-alternatives><mixed-citation xml:lang="ru">Brandenburg J.M., Müller-Stich B.P., Wagner M., van der Schaar M. Can surgeons trust AI? Perspectives on machine learning in surgery and the importance of eXplainable Artificial Intelligence (XAI) // Langenbeck’s Archives of Surgery. 2025. V. 410. N 1. P. 53. https://doi.org/10.1007/s00423-025-03626-7</mixed-citation><mixed-citation xml:lang="en">Brandenburg J.M., Müller-Stich B.P., Wagner M., van der Schaar M. Can surgeons trust AI? Perspectives on machine learning in surgery and the importance of eXplainable Artificial Intelligence (XAI) // Langenbeck’s Archives of Surgery. 2025. V. 410. N 1. P. 53. https://doi.org/10.1007/s00423-025-03626-7</mixed-citation></citation-alternatives></ref><ref id="cit4"><label>4</label><citation-alternatives><mixed-citation xml:lang="ru">Kalasampath K., Spoorthi K.N., Sajeev S., Kuppa S.S., Ajay K., Maruthamuthu A. A Literature review on applications of explainable artificial intelligence (XAI) // IEEE Access. 2025. V. 13. P. 41111–41140. https://doi.org/10.1109/access.2025.3546681</mixed-citation><mixed-citation xml:lang="en">Kalasampath K., Spoorthi K.N., Sajeev S., Kuppa S.S., Ajay K., Maruthamuthu A. A Literature review on applications of explainable artificial intelligence (XAI) // IEEE Access. 2025. V. 13. P. 41111–41140. https://doi.org/10.1109/access.2025.3546681</mixed-citation></citation-alternatives></ref><ref id="cit5"><label>5</label><citation-alternatives><mixed-citation xml:lang="ru">Kalmykov V.L., Kalmykov L.V. Towards eXplicitly eXplainable Artificial Intelligence // Information Fusion. 2025. V. 123. P. 103352. https://doi.org/10.1016/j.inffus.2025.103352</mixed-citation><mixed-citation xml:lang="en">Kalmykov V.L., Kalmykov L.V. Towards eXplicitly eXplainable Artificial Intelligence // Information Fusion. 2025. V. 123. P. 103352. https://doi.org/10.1016/j.inffus.2025.103352</mixed-citation></citation-alternatives></ref><ref id="cit6"><label>6</label><citation-alternatives><mixed-citation xml:lang="ru">Mohanty P.K., Francis S.A.J., Barik R.K., Reddy K.H.K., Roy D.S., Saikia M.J. IMPACT: an interactive multi-disease prevention and counterfactual treatment system using explainable AI and a multimodal LLM // PeerJ Computer Science. 2025. V. 11. P. e2839. https://doi.org/10.7717/peerj-cs.2839</mixed-citation><mixed-citation xml:lang="en">Mohanty P.K., Francis S.A.J., Barik R.K., Reddy K.H.K., Roy D.S., Saikia M.J. IMPACT: an interactive multi-disease prevention and counterfactual treatment system using explainable AI and a multimodal LLM // PeerJ Computer Science. 2025. V. 11. P. e2839. https://doi.org/10.7717/peerj-cs.2839</mixed-citation></citation-alternatives></ref><ref id="cit7"><label>7</label><citation-alternatives><mixed-citation xml:lang="ru">Gadekallu T.R., Yenduri G., Kaluri R., Rajput D.S., Lakshmanna K., Fang K., Chen J.X., Wang W. The role of GPT in promoting inclusive higher education for people with various learning disabilities: a review // PeerJ Computer Science. 2025. V. 11. P. e2400. https://doi.org/10.7717/peerj-cs.2400</mixed-citation><mixed-citation xml:lang="en">Gadekallu T.R., Yenduri G., Kaluri R., Rajput D.S., Lakshmanna K., Fang K., Chen J.X., Wang W. The role of GPT in promoting inclusive higher education for people with various learning disabilities: a review // PeerJ Computer Science. 2025. V. 11. P. e2400. https://doi.org/10.7717/peerj-cs.2400</mixed-citation></citation-alternatives></ref><ref id="cit8"><label>8</label><citation-alternatives><mixed-citation xml:lang="ru">Aydin O., Karaarslan E., Erenay F.S., Bacaninet N. Generative AI in academic writing: A comparison of DeepSeek, Qwen, ChatGPT, Gemini, Llama, Mistral, and Gemma // arXiv. 2025. arXiv:2503.04765. https://doi.org/10.48550/arXiv.2503.04765</mixed-citation><mixed-citation xml:lang="en">Aydin O., Karaarslan E., Erenay F.S., Bacaninet N. Generative AI in academic writing: A comparison of DeepSeek, Qwen, ChatGPT, Gemini, Llama, Mistral, and Gemma // arXiv. 2025. arXiv:2503.04765. https://doi.org/10.48550/arXiv.2503.04765</mixed-citation></citation-alternatives></ref><ref id="cit9"><label>9</label><citation-alternatives><mixed-citation xml:lang="ru">Kamath A., Ferret J., Pathak S., Vieillard N., Merhej R., Perrin S., et al. Gemma 3 technical report // arXiv. 2025. arXiv:2503.19786. https://doi.org/10.48550/arXiv.2503.19786</mixed-citation><mixed-citation xml:lang="en">Kamath A., Ferret J., Pathak S., Vieillard N., Merhej R., Perrin S., et al. Gemma 3 technical report // arXiv. 2025. arXiv:2503.19786. https://doi.org/10.48550/arXiv.2503.19786</mixed-citation></citation-alternatives></ref><ref id="cit10"><label>10</label><citation-alternatives><mixed-citation xml:lang="ru">Dwivedi R., Dave D., Naik H., Singhal S., Omer R., Patel P., et al. Explainable AI (XAI): core ideas, techniques, and solutions // ACM Computing Surveys. 2023. V. 55. N 9. P. 194. https://doi.org/10.1145/3561048</mixed-citation><mixed-citation xml:lang="en">Dwivedi R., Dave D., Naik H., Singhal S., Omer R., Patel P., et al. Explainable AI (XAI): core ideas, techniques, and solutions // ACM Computing Surveys. 2023. V. 55. N 9. P. 194. https://doi.org/10.1145/3561048</mixed-citation></citation-alternatives></ref><ref id="cit11"><label>11</label><citation-alternatives><mixed-citation xml:lang="ru">Salih A.M., Raisi-Estabragh Z., Galazzo I.B., Radeva P., Petersen S.E., Lekadir K., Menegaz G. A perspective on explainable artificial intelligence methods: SHAP and LIME // Advanced Intelligent Systems. 2025. V. 7. N 1. P. 2400304. https://doi.org/10.1002/aisy.202400304</mixed-citation><mixed-citation xml:lang="en">Salih A.M., Raisi-Estabragh Z., Galazzo I.B., Radeva P., Petersen S.E., Lekadir K., Menegaz G. A perspective on explainable artificial intelligence methods: SHAP and LIME // Advanced Intelligent Systems. 2025. V. 7. N 1. P. 2400304. https://doi.org/10.1002/aisy.202400304</mixed-citation></citation-alternatives></ref><ref id="cit12"><label>12</label><citation-alternatives><mixed-citation xml:lang="ru">Mersha M.A., Yigezu M.G., Tonja A.L., Shakil H., Iskander S., Kolesnikova O., Kalita J. Explainable AI: XAI-guided context-aware data augmentation // Expert Systems with Applications. 2025. V. 289. P. 128364. https://doi.org/10.1016/j.eswa.2025.128364</mixed-citation><mixed-citation xml:lang="en">Mersha M.A., Yigezu M.G., Tonja A.L., Shakil H., Iskander S., Kolesnikova O., Kalita J. Explainable AI: XAI-guided context-aware data augmentation // Expert Systems with Applications. 2025. V. 289. P. 128364. https://doi.org/10.1016/j.eswa.2025.128364</mixed-citation></citation-alternatives></ref><ref id="cit13"><label>13</label><citation-alternatives><mixed-citation xml:lang="ru">Ng K.K.Y., Matsuba I., Zhang P.C. RAG in health care: a novel framework for improving communication and decision-making by addressing LLM limitations // NEJM AI. 2025. V. 2. N 1. P. 2400380. https://doi.org/10.1056/AIra2400380</mixed-citation><mixed-citation xml:lang="en">Ng K.K.Y., Matsuba I., Zhang P.C. RAG in health care: a novel framework for improving communication and decision-making by addressing LLM limitations // NEJM AI. 2025. V. 2. N 1. P. 2400380. https://doi.org/10.1056/AIra2400380</mixed-citation></citation-alternatives></ref><ref id="cit14"><label>14</label><citation-alternatives><mixed-citation xml:lang="ru">Strom B.E., Applebaum A., Miller D.P., Nickels K.C., Pennington A.G., Thomas C.B. MITRE ATT&amp;C</mixed-citation><mixed-citation xml:lang="en">Strom B.E., Applebaum A., Miller D.P., Nickels K.C., Pennington A.G., Thomas C.B. MITRE ATT&amp;C</mixed-citation></citation-alternatives></ref></ref-list><fn-group><fn fn-type="conflict"><p>The authors declare that there are no conflicts of interest present.</p></fn></fn-group></back></article>
