kaz29Вестник Казахстанско-Британского технического университетаHerald of the Kazakh-British Technical University1998-66882959-8109Казахстанско-Британский Технический Университет10.55452/1998-6688-2026-23-1-173-184kaz29-2511Research ArticleКОМПЬЮТЕРНЫЕ НАУКИCOMPUTER SCIENCEОБНАРУЖЕНИЕ МОШЕННИЧЕСКИХ ТЕЛЕФОННЫХ ЗВОНКОВ В РЕАЛЬНОМ ВРЕМЕНИ С ИСПОЛЬЗОВАНИЕМ МНОГОХОДОВОГО АНАЛИЗА ДИАЛОГАREAL-TIME DETECTION OF FRAUDULENT PHONE CALLS USING MULTI-TURN DIALOGUE ANALYSIShttps://orcid.org/0000-0001-7096-6765СерекА.SerekA.,

PhD, ассоциированный профессор

г. Астана

PhD, Associate Professor

Astana

Azamat.Serek@astanait.edu.kzhttps://orcid.org/0000-0002-9328-8300ШойынбекА.ShoiynbekA.

PhD, профессор

г. Алматы

PhD, Professor

Almaty

aisultan.shoiynbek@narxoz.kzhttps://orcid.org/0000-0001-5952-8609КуанышбайД.KuanyshbayD.

PhD, ассистент-профессор

г. Алматы

PhD, Assistant Professor

Almaty

darkhan.kuanyshbay@narxoz.kzAstana IT университетКазахстанAstana IT UniversityKazakhstanNarxoz университетКазахстанNarxoz UniversityKazakhstan202629032026231173184Copyright © Серек А., Шойынбек А., Куанышбай Д., 20262026Серек А., Шойынбек А., Куанышбай Д.Serek A., Shoiynbek A., Kuanyshbay D.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://vestnik.kbtu.edu.kz/jour/article/view/2511

Расширение телекоммуникационных услуг сопровождалось ростом числа мошеннических телефонных звонков, представляющих серьезную угрозу для отдельных лиц и организаций. Традиционные методы обнаружения обычно основаны на офлайн-анализе полных разговоров, что ограничивает оперативность реагирования.В данной работе автор предлагает систему обнаружения мошенничества в реальном времени, основанную на предварительно обученных контекстных встраиваниях в сочетании с двунаправленной сетью долговременной кратковременной памяти (LSTM) для моделирования семантического содержания и временной динамики многоходовых разговоров. Для обнаружения мошеннических звонков система постепенно изменяет вероятность мошенничества после каждого хода разговора, что позволяет обнаружить мошенничество. При тестировании на синтетическом наборе данных многоходовых диалогов показано, что предложенная BiLSTM с использованием встраиваний BERT имеет точность 93,75% и показатель F1 93,55, что выше, чем у существующих базовых моделей машинного обучения и сверточных нейронных сетей. Система способна выявлять большинство мошеннических схем на начальных этапах звонка, что обеспечивает быструю оценку риска. Эти результаты свидетельствуют о полезности контекстно ориентированного, последовательного моделирования для обнаружения мошенничества в режиме реального времени и о возможности его практического применения.

The expansion of telecommunication services has been met with a rise in the cases of the fraudulent phone calls posing a big threat to individuals and organizations. Conventional techniques of detecting are usually based on offline analysis of full conversations, which restricts their promptness of intervention. In this paper, the author proposes a real-time, turn-taking, fraud detecting system, which is based on pre-trained contextual embeddings in combination with a bi-directional Long Short-Term Memory network in order to model semantic content and temporal dynamics of multi-turn conversations. To detect fraudulent calls, the system progressively changes the probability of a call being a fraud after every conversational turn to allow it to detect a fraud. When tested with a synthetic multi-turn dialogue dataset, it is shown that the proposed BiLSTM using BERT embeddings has a test accuracy of 93.75% and an F 1 score of 93.55, which is higher than the current machine learning and convolutional baselines. The system can note most of the scams during the initial few turns of a call, which offers fast risk evaluation. These findings suggest the usefulness of context-based, progressing modeling to detect fraud in real time and its possibility of practical application.

обнаружение мошенничества в реальном временителефонные аферымногоходовый диалогBiLSTMконтекстные вложенияпоследовательное моделированиераннее вмешательствотелекоммуникационная безопасностьпрогнозирование уровня хода диалогапотоковый анализreal-time fraud detectionphone scamsmulti-turn dialogueBiLSTMcontextual embeddingssequential modelingearly interventiontelecommunication securityturn-level predictionstreaming analysisThis research has been funded by the Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (Grant No. AP27510301 “Development of technology for recognizing fraudulent actions during a telephone conversation and/or text message exchange in messengers based on artificial intelligence algorithms”).ReferencesSergeevna, P.E., Timurovich, G.S., and Gennadievich, B.P. The factor of complex interaction in responding to telephone fraud. Voprosy Bezopasnosti, 1, 1–9 (2023). https://doi.org/10.25136/2409-7543.2023.1.39274Sergeevna, P.E., Timurovich, G.S., and Gennadievich, B.P. The factor of complex interaction in responding to telephone fraud. Voprosy Bezopasnosti, 1, 1–9 (2023). https://doi.org/10.25136/2409-7543.2023.1.39274Syafitri, W., Shukur, Z., Mokhtar, U.A., Sulaiman, R., and Ibrahim, M.A. Social engineering attacks prevention: A systematic literature review. IEEE Access, 10, 39325–39343 (2022). https://doi.org/10.1109/ACCESS.2022.3162594Syafitri, W., Shukur, Z., Mokhtar, U.A., Sulaiman, R., and Ibrahim, M.A. Social engineering attacks prevention: A systematic literature review. IEEE Access, 10, 39325–39343 (2022). https://doi.org/10.1109/ACCESS.2022.3162594Anuar, H.A., et al. Phone scam unveiled: Insights from a systematic literature review. Journal of Financial Crime (2025). https://doi.org/10.1108/JFC-03-2025-0078Anuar, H.A., et al. Phone scam unveiled: Insights from a systematic literature review. Journal of Financial Crime (2025). https://doi.org/10.1108/JFC-03-2025-0078Soudani, H., et al. A survey on recent advances in conversational data generation. ACM Computing Surveys (2024). https://doi.org/10.1145/3795686Soudani, H., et al. A survey on recent advances in conversational data generation. ACM Computing Surveys (2024). https://doi.org/10.1145/3795686Kusal, S., Patil, S., Choudrie, J., Kotecha, K., Mishra, S., and Abraham, A. AI-based conversational agents: A scoping review from technologies to future directions. IEEE Access, 10, 92337–92356 (2022). https://doi.org/10.1109/ACCESS.2022.3201144Kusal, S., Patil, S., Choudrie, J., Kotecha, K., Mishra, S., and Abraham, A. AI-based conversational agents: A scoping review from technologies to future directions. IEEE Access, 10, 92337–92356 (2022). https://doi.org/10.1109/ACCESS.2022.3201144Singh, S., and Beniwal, H. A survey on near-human conversational agents. Journal of King Saud University – Computer and Information Sciences, 34 (10), 8852–8866 (2022). https://doi.org/10.1016/j.jksuci.2021.10.013Singh, S., and Beniwal, H. A survey on near-human conversational agents. Journal of King Saud University – Computer and Information Sciences, 34 (10), 8852–8866 (2022). https://doi.org/10.1016/j.jksuci.2021.10.013Rim, D., et al. To chat or task: A multi-turn dialogue generation framework for task-oriented dialogue systems. In: Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Industry Track), 576–592 (2025). https://doi.org/10.18653/v1/2025.acl-industry.41Rim, D., et al. To chat or task: A multi-turn dialogue generation framework for task-oriented dialogue systems. In: Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Industry Track), 576–592 (2025). https://doi.org/10.18653/v1/2025.acl-industry.41Pattnayak, P., et al. Hybrid AI for responsive multi-turn online conversations with novel dynamic routing and feedback adaptation. In: Proceedings of the 4th International Workshop on Knowledge-Augmented Methods for Natural Language Processing, 215–229 (2025). https://doi.org/10.18653/v1/2025.knowledgenlp-1.20Pattnayak, P., et al. Hybrid AI for responsive multi-turn online conversations with novel dynamic routing and feedback adaptation. In: Proceedings of the 4th International Workshop on Knowledge-Augmented Methods for Natural Language Processing, 215–229 (2025). https://doi.org/10.18653/v1/2025.knowledgenlp-1.20Li, X., et al. Proactive guidance of multi-turn conversation in industrial search. In: Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Industry Track), 706–717 (2025). https://doi.org/10.18653/v1/2025.acl-industry.50Li, X., et al. Proactive guidance of multi-turn conversation in industrial search. In: Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Industry Track), 706–717 (2025). https://doi.org/10.18653/v1/2025.acl-industry.50Pandey, A. Retrieval augmented fraud detection. In: Proceedings of the 5th ACM International Conference on AI in Finance, 328–335 (2024). https://doi.org/10.1145/3677052.3698692Pandey, A. Retrieval augmented fraud detection. In: Proceedings of the 5th ACM International Conference on AI in Finance, 328–335 (2024). https://doi.org/10.1145/3677052.3698692Perera, L., et al. AE-RAGX: Combining autoencoders with retrieval-augmented generation for explainable anomaly detection using LLMs. In: 2025 IEEE Latin-American Conference on Communications (LATINCOM), 1–6 (2025). https://doi.org/10.1109/LATINCOM67778.2025.11345384Perera, L., et al. AE-RAGX: Combining autoencoders with retrieval-augmented generation for explainable anomaly detection using LLMs. In: 2025 IEEE Latin-American Conference on Communications (LATINCOM), 1–6 (2025). https://doi.org/10.1109/LATINCOM67778.2025.11345384Xu, J. Enhancing financial risk management with retrieval-augmented large language models. In: Proceedings of the 4th International Conference on Artificial Intelligence, Internet and Digital Economy (ICAID), 138–141 (2025). https://doi.org/10.1109/ICAID65275.2025.11034536Xu, J. Enhancing financial risk management with retrieval-augmented large language models. In: Proceedings of the 4th International Conference on Artificial Intelligence, Internet and Digital Economy (ICAID), 138–141 (2025). https://doi.org/10.1109/ICAID65275.2025.11034536Li, Z. Knowledge-grounded detection of cryptocurrency scams with retrieval-augmented LMs. In: Proceedings of the 3rd Workshop on Towards Knowledgeable Foundation Models (KnowFM), 40–48 (2025). https://doi.org/10.18653/v1/2025.knowllm-1.4Li, Z. Knowledge-grounded detection of cryptocurrency scams with retrieval-augmented LMs. In: Proceedings of the 3rd Workshop on Towards Knowledgeable Foundation Models (KnowFM), 40–48 (2025). https://doi.org/10.18653/v1/2025.knowllm-1.4Chang, Y.C. Scam detection with large language models: Multimodal risk analysis of URLs and chat messages (2025). https://doi.org/10.71781/239Chang, Y.C. Scam detection with large language models: Multimodal risk analysis of URLs and chat messages (2025). https://doi.org/10.71781/239Akbar, K.A., et al. Retrieval augmented generation-based large language models for bridging transportation cybersecurity legal knowledge gaps. Transportation Research Record (2025). https://doi.org/10.1177/03611981251372471Akbar, K.A., et al. Retrieval augmented generation-based large language models for bridging transportation cybersecurity legal knowledge gaps. Transportation Research Record (2025). https://doi.org/10.1177/03611981251372471Malhotra, S., Arora, G., and Bathla, R. Detection and analysis of fraud phone calls using artificial intelligence. In: Proceedings of the International Conference on Recent Advances in Electrical, Electronics & Digital Healthcare Technologies (REEDCON), 592–595 (2023). https://doi.org/10.1109/REEDCON57544.2023.10150631Malhotra, S., Arora, G., and Bathla, R. Detection and analysis of fraud phone calls using artificial intelligence. In: Proceedings of the International Conference on Recent Advances in Electrical, Electronics & Digital Healthcare Technologies (REEDCON), 592–595 (2023). https://doi.org/10.1109/REEDCON57544.2023.10150631Cazzolato, M., et al. CallMine: Fraud detection and visualization of million-scale call graphs. In: Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, 4509–4515 (2023). https://doi.org/10.1145/3583780.3614662Cazzolato, M., et al. CallMine: Fraud detection and visualization of million-scale call graphs. In: Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, 4509–4515 (2023). https://doi.org/10.1145/3583780.3614662Wahid, A., et al. NFA: A neural factorization autoencoder based online telephony fraud detection. Digital Communications and Networks, 10 (1), 158–167 (2024). https://doi.org/10.1016/j.dcan.2023.03.002Wahid, A., et al. NFA: A neural factorization autoencoder based online telephony fraud detection. Digital Communications and Networks, 10 (1), 158–167 (2024). https://doi.org/10.1016/j.dcan.2023.03.002Ren, L., et al. Dynamic graph neural network-based fraud detectors against collaborative fraudsters. Knowledge-Based Systems, 278, 110888 (2023). https://doi.org/10.1016/j.knosys.2023.110888Ren, L., et al. Dynamic graph neural network-based fraud detectors against collaborative fraudsters. Knowledge-Based Systems, 278, 110888 (2023). https://doi.org/10.1016/j.knosys.2023.110888Wu, Y., et al. Fraud-agents detection in online microfinance: A large-scale empirical study. IEEE Transactions on Dependable and Secure Computing, 20 (2), 1169–1185 (2022). https://doi.org/10.1109/TDSC.2022.3151132Wu, Y., et al. Fraud-agents detection in online microfinance: A large-scale empirical study. IEEE Transactions on Dependable and Secure Computing, 20 (2), 1169–1185 (2022). https://doi.org/10.1109/TDSC.2022.3151132

The authors declare that there are no conflicts of interest present.