The publication provides brief information about the VII All–Russian Conference with international participation " VII All-Russian Conference with International Participation «Digital Technologies of the Future — Modern Solutions in Earth Sciences. ITES-2025", which took place on September 22–26, 2025 in Vladivostok.

This paper examines the microseismic noise spectra a few hours before moderate and strong seismic events. Forty earthquakes with an energy class of K=9.5–14.5 at epicentral distances of 10 to 120 km were considered. A statistically significant increase in the spectral power density (SPD) was detected in the 0.8–2.4 Hz range. Machine learning methods were used to construct a binary classification model that allows detection of earthquake preparations a few hours before an event based on microseismic SPD values in the specified frequency range.

The article presents an innovative methodology for applying Semantic Web technologies to support fundamental geological research. The problem of semantic integration of heterogeneous geological data, characterized by different scales and interdisciplinarity, is considered. A five-stage methodology is developed, including domain analysis, ontological conceptual modeling, data transformation into a knowledge graph, deployment of a distributed data access infrastructure based on the conceptual model, and integration with processing and analysis procedures. Practical testing was conducted on three case studies: analysis of geochemical data for assessing territory pollution levels, creation of an information system about faults, and research on reservoir shoreline dynamics. The proposed ontological approach ensures compliance with FAIR principles and overcoming the "semantic barrier" in geological research. It is shown that Semantic Web technologies enable a transition from fragmented information arrays to a holistic semantic space of geological knowledge, opening new opportunities for generating comprehensive scientific hypotheses and cross-disciplinary research.

This article presents the concept and architecture of a multi-agent system designed to function as a digital assistant for geologist researchers. The system aims to automate key stages of scientific research: from topic formulation and literature review to hypothesis generation and presentation of results. The article describes the system's integration with the GeologyScience.ru platform, which provides access to diverse geological data and analysis tools, as well as approaches to adapting large-scale language models (LLM) to solve specialized scientific problems.

The article discusses technologies that can radically change the development of many areas at once: artificial intelligence, quantum technologies, big data, wireless communication technologies, distributed registry systems. The authors consider a number of promising technologies of the near future that currently have prospects for application in Earth sciences. The review of the application of these technologies to solve various geological problems, including the results obtained by the authors, is carried out.

The rapid increase in the volume of publications in various fields of geology makes it crucial to introduce methods for automated processing of scientific texts. Large language models based on neural networks represent one of the most promising approaches to solving this challenge. The recent breakthroughs in artificial intelligence have made such models indispensable tools for researchers. Our work on semantic search for publications using additionally trained language models and measuring the similarity between geological texts yielded good results. However, the models we used were unable to perform in-depth text analysis. A comparative analysis of modern architectures identified the DeepSeek R1 model as belonging to a class of systems with advanced logical inference abilities. This type of model represents a fundamentally new level of quality in text generation. Based on the chosen model, we have developed a web service that provides unique functionality for comparative analysis of up to 5 scientific articles. The service supports multilingual sources, allowing users to input text in English, Chinese, Russian, etc. It generates structured reports in Russian, highlighting key theses, contradictions, and patterns. The proposed approach has been tested on geological publications, and the results have been promising.

The experience of the development of digital technologies in geology and extractive industries, including governmental and private structures, industrial, scientific and educational facilities, is considered. The results of the using of digital technologies for a wide range of studies of geological exploration and production, the advantages and disadvantages are considered. The approaches that can provide a qualitative increase in knowledge and information in the field of geology are proposed.

A comparative analysis of image segmentation methods for fire detection was conducted using thresholding in the HSV color space and the U-Net neural network. The study aimed to evaluate the efficiency of these approaches in terms of execution time and fire detection accuracy based on RMSE, IoU, Dice, and MAPE metrics. Experiments were performed on four different fire images with manually prepared ground truth fire masks. The results showed that the HSV method offers high processing speed (0.0010–0.0020 s) but tends to detect not only fire but also smoke, reducing its accuracy (IoU 0.0863–0.3357, Dice 0.1588–0.5026). The U-Net neural network demonstrates higher fire segmentation accuracy (IoU up to 0.6015, Dice up to 0.7512) due to selective flame detection but requires significantly more time (1.2477–1.3733 s) and may underestimate the total fire area (MAPE up to 78.5840%). Visual assessment confirmed differences in methods' behavior: HSV captures smoke as part of the target area, while U-Net focuses exclusively on fire. The choice between methods depends on task priorities: speed or accuracy. Future research directions were proposed, including U-Net optimization and the development of hybrid approaches.

In this paper, we study the inverse problem of identifying the dimensionless thermal conductivity coefficient for the Green–Naghdi equation of type III, which describes the propagation of thermal disturbances with a finite velocity and takes into account the inertial effects of heat flux. For the inverse problem, the stability requirement (Hadamard criteria) is violated, as a result of which even minimal data distortions lead to significant errors in parameter identification. As a solution method, we use an approach based on physically informed neural networks (PINN), which combines the capabilities of deep learning with a priori knowledge of the structure of the differential equation. The parameter is included among the trained variables, and the loss function is formed based on the deviation from the differential equation, boundary conditions, initial conditions, and noisy experimental data from a point sensor. The results of computational experiments are presented, demonstrating high accuracy of parameter recovery (error less than 0.03%) and the stability of the method with respect to the presence of additive Gaussian noise in the data. The PINN method has proven itself to be an effective tool for solving ill-posed inverse problems of mathematical physics.

Methods of kernel estimation of a priori density in the deconvolution problem are used to construct guaranteed procedures for distinguishing between two one-sided hypotheses. The situation is considered when the observed random variable is the sum of an unknown parameter and a centered normal error with a known variance. Consistent empirical estimates are constructed for the d-posterior risk function. The convergence of the corresponding critical constant to the optimal value is established. The accuracy of the procedures is illustrated numerically on three variants of the prior distribution.

Increasing urbanization and motorization lead to a shortage of parking spaces, resulting in congestion, increased emissions, and a declining quality of life. Traditional parking management methods are ineffective in addressing this issue, necessitating the use of data analysis and forecasting tools. This paper examines the use of a digital twin of the Kazan parking system. Data was filtered and integrated, points of interest were clustered, and a correlation analysis of factors influencing parking occupancy was performed. Linear regression, decision tree, random forest, XGBoost, MLP, and LSTM models were trained and compared to predict occupancy levels. The random forest model demonstrated the best results. The developed digital twin prototype enables monitoring and scenario modeling, making it an effective tool for parking space optimization and management decision-making.

This paper presents the development of Tula, a tool for video game balancing. The necessity for such a tool is substantiated by the growing requirements for quality and cost-effectiveness in the video game industry, particularly in managing in-game economy and game world logic. The study analyzes existing tools and approaches to game balancing, identifying their limitations, which informed the design of the new tool's functionality. The presented tool integrates features of contemporary solutions while providing enhanced capabilities for game parameter analysis and testing, including prototype generation via class descriptions and real-time simulation. The technological foundation and architecture of the tool are described in detail. Key implementation aspects are discussed: interface responsiveness, continuous data synchronization, and security. Comparative analysis with Machinations revealed advantages in data processing correctness, interface convenience, and prototype modification flexibility.

A new precise MCE algorithm for extracting the main content from news websites is presented. The proposed algorithm uses analysis of the Document Object Model (DOM) structure and content density metrics to identify and extract the informational core of a web page. The implemented approach combines three key features: the maximum number of direct child elements containing text, the maximum textual content without child elements containing text, and the closest position to the average node depth. The algorithm demonstrated superior performance compared to existing solutions such as Boilerpipe and Readability, achieving 99.96% precision, 99.69% recall, and 99.80% F1-score on a comprehensive dataset of 500 diverse web pages. Its language-independent design makes the algorithm particularly effective for extracting multilingual content, including languages with complex structures such as Arabic.

The paper considers a method for classifying unmanned aerial vehicles (UAVs) and birds based on radar measurements in various object movement scenarios. The relevance of the problem is explained by the complexity of UAV detection due to its small overall dimensions, high agility, and similar geometry to a bird. To solve the problem, a classification method based on the analysis of micro-Doppler signatures, which reflect the dynamics of object motion, is applied. The scenarios are modelled in the Engee environment, where models of radar system, UAV and birds are developed.

This article addresses the problem of creating dynamic narrative systems for video games with real-time interactivity. It presents the development and testing of a GPT integration component for dialogue generation, which revealed a critical limitation of cloud-based solutions – a 30-second latency unacceptable for gameplay. A hybrid architecture of an adaptive system is proposed, combining LLMs with reinforcement learning mechanisms. Particular attention is given to solving the problems of game world consistency and managing long-term context of NPC interactions through a RAG approach. The transition to the Edge AI paradigm with the application of quantization methods to achieve a target latency of 200–500 ms is substantiated. Metrics for evaluating personalization and dynamic content adaptation have been developed.