UDC 65.012.76
The possibility of building a neural network classifier of texts describing emergency situations of various types is considered. The basis is the construction of a multi-layer neural network for training on pre-lemmatized texts using the tf-idf algorithm. The network is trained to recognize the description of nine emergency situations and, after their classification, to attach the necessary materials, from regulations to precedents. The classification reliability is 0.89-0.90. The classifier has two additional modules: a module for parsing information sources (telegram channels and news resources), and a module for geoparsing, which allows you to display a map of the incident location and its geographical coordinates, as well as perform statistical and cluster analysis of the data, if necessary. A method is proposed to increase the productivity of officials and the efficiency of field response units. The topology of a neural network is defined as a neural network model for classifying emergency situations, and a mechanism for parsing and processing information is developed. A comparison is made with well-known systems such as ChatGPT and DeepSeek. The results show better accuracy in predictions and take into account information security requirements. Practical application will reduce the time it takes to process information about fires and emergencies
neural network, text classifier, emergency, management, data parsing
1. Jones K.S. A statistical interpretation of term specificity and its application in retrieval (angl.) // Journal of Documentation: zhurnal. - MSVUniversity: MCB University Press, 2004. - Vol. 60, no. 5. - P. 493 - 502. - ISSN 0022-0418.
2. Avdeenko A.M., Mel'nichenko A.S., Habibulin R.Sh. Vvedenie v teoriyu intellektual'nogo analiza dannyh. - M.: Akademiya GPS MChS Rossii, 2018. - 143 s.
3. Keras. Built-in loss functions. [Elektronnyy resurs] URL: https://github.com/keras-team/keras/blob/master/keras/losses.py (data obrascheniya: 27.06.2025).
4. Tensorflow/python/keras/backend.py. [Elektronnyy resurs] URL: https://github.com/tensorflow/tensorflow/blob/r1.12/tensorflow/python/keras/backend.py (data obrascheniya: 29.06.2025).
5. Volkov N.A., Ivanov A.V. Podhod k ocenke zaschischennosti rechevoy akusticheskoy informacii s primeneniem neyronnyh setey // Vestnik SibGUTI. 2024. T. 18, No 2. S. 43-56. DOI:https://doi.org/10.55648/1998-6920-2024-18-2-43-56 EDN: https://elibrary.ru/EKGZTF
6. Voronina I.E., Pastrevich M.K. Model' klassifikacii verbal'noy agressii v nestrukturirovannyh tekstah // Vestnik SibGUTI. 2025. T. 19, No 2. S. 81-88. DOI:https://doi.org/10.55648/1998-6920-2025-19-2-81-88 EDN: https://elibrary.ru/EIBXTG
7. Percev I.V., Sitnyakovskaya E.I. Lingvisticheskoe obespechenie informacionnyh sistem i processov // Vestnik SibGUTI. 2024. T. 18, No 1. S. 70-77. DOI:https://doi.org/10.55648/1998-6920-2024-18-1-70-77 EDN: https://elibrary.ru/HIEJZG
8. L. Bourdev and J. Malik. Poselets: Body part detectors trained using 3d human pose annotations. In International Conference on Computer Vision (ICCV), 2009. DOI:https://doi.org/10.1109/ICCV.2009.5459303
9. H. Cai, Q. Wu, T. Corradi, and P. Hall. The crossdepiction problem: Computer vision algorithms for recognising objects in artwork and in photographs. arXiv preprint arXiv:1505.00110,2015.. DOI:https://doi.org/10.48550/arXiv.1505.00110
10. T. Dean, M. Ruzon, M. Segal, J. Shlens, S. Vijayanarasimhan, J. Yagnik, et al. Fast, accurate detection of 100,000 object classes on a single machine. In Computer Vision and Pattern Recognition (CVPR), 2013 IEEE Conference on, pages 1814-1821. IEEE, 2013. DOI:https://doi.org/10.1109/CVPR.2013.237
11. J. Donahue, Y. Jia, O. Vinyals, J. Hoffman, N. Zhang, E. Tzeng, and T. Darrell. Decaf: A deep convolutional activation feature for generic visual recognition. arXiv preprint arXiv:1310.1531, 2013. DOI:https://doi.org/10.48550/arXiv.1310.1531
12. J. Dong, Q. Chen, S. Yan, and A. Yuille. Towards unified object detection and semantic segmentation. In Computer Vision-ECCV 2014, pages 299-314.Springer, 2019. DOI:https://doi.org/10.1007/978-3-319-10602-1_20
13. A.M. Avdeenko, A.P. Satin, E.V. Gavrilyuk. Podderzhka prinyatiya upravlencheskih resheniy na osnove neyrosetevyh klassifikatorov tekstov/"Informatika: problemy, metody, tehnologii" (IPMT), Novosibirsk 7-10.2, 2024 g. EDN: https://elibrary.ru/CFFWWK
14. Prikaz MChS Rossii ot 28 dekabrya 2022 g. No 1333 "Ob utverzhdenii Poryadka organizacii deyatel'nosti professional'nyh avariyno-spasatel'nyh formirovaniy MChS Rossii, vypolnyayuschih poiskovo-spasatel'nye raboty".
15. Elektronnyy resurs https://chat.deepseek.com/(data obrascheniya: 18.07.2025).
16. Elektronnyy resurs https://chatgpt.com/(data obrascheniya: 18.07.2025).
17. ChatGPT and Other Large Language Models Are Double-edged Swords / Y. Shen [et.al.] // New York University, Center for Data Science, 2023 - P. 5-9.
18. Gradyushko, A.A. Chat-bot v Telegram kak novyy istochnik informacii dlya zhurnalistov / A. A. Gradyushko. - Tekst: neposredstvennyy // MEDIAObrazovanie: media 80 kak total'naya povsednevnost': materialy V Mezhdunarodnoy nauchnoy konferencii (24 25 noyabrya 2020 goda, Chelyabinsk). / pod red. A.A. Morozovoy. - Chelyabinsk: Izdatel'stvo Chelyabinskogo gosudarstvennogo universiteta, 2020. - S. 305-309. EDN: https://elibrary.ru/KLBAIT
19. Frank A., Asuncion A. UCLMachineLearn ingRepository. University of California, School of Information and Computer Science. Irvine, 2010. [Elektron. resurs] Rezhim dostupa: http://arhive.ics.uci.edu/ml (data obrascheniya: 17.07.2025).
20. A.V. Bahtizin, V.V. Korablev, A.N. Morozov i dr. Iskusstvennyy intellekt v neftegazovoy promyshlennosti: problemy i perspektivy /// Neftegazovoe delo. 2019. T. 17. No 1. S. 5-11.
21. Abu-Abed F.N. "Postroenie neyrosetevogo klassifikatora dlya obnaruzheniya ne shtatnyh situaciy v processe promyshlennogo bureniya neftyanyh skvazhin". TERRITORIYa NEFTEGAZ // Burenie //vypusk No 6 // iyun' - 2012. - Moskva, 2012. - S. 16-19.
22. Kosinski M., Stillwell D., Graepel T. Private traits and attributes are predictable from digital records of human behavior // Proceedings of the National Academy of Sciences (PNAS). 2013. Vol. 110, N 15. P. 5802 5805.
