Abstract:

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.

Abstract:

Starting from spring 2022 a national system for evaluating the effectiveness of scientific research and development is being created using the database of leading Russian journals Russian Science Citation Index (RSCI). The article represents some results of the analysis of the RSCI normalized rating published in December 2022. Using the journal citation graph constructed from the data Math-Net.Ru on the example of thematic group OECD 101.Mathematics we show that in the case of a large number of self-citations of the journals, a relationship between the normalized rating and the number of references is found, and the relationship with self-citation is somewhat stronger than without self-citation. Analysis of the distribution of the journals by thematic groups shows that using a single attribute, such as the OECD, as a criterion does not allow forming a group as a community of journals with a single strongly connected component in the citation graph. We conclude that the research methods of journal citation graphs are a good basis for a comparative analysis of the characteristics of journals and their ranking, which means they can be used as tools for further development and improvement of journal ratings.

Abstract:

Human Pose Estimation (HPE) has become one of the most relevant topics in computer vision research. This technology can be applied in various fields such as video surveillance, medical care, and sports motion analysis. Due to the increasing demand for HPE, many libraries for this technology have been developed in the last 20 years. Since 2017, many HPE algorithms based on skeletal model have been published and packaged into libraries for easy use by researchers.

These libraries are important for researchers who want to integrate them into real-world applications for video surveillance, medical care, and sports motion analysis.

This paper investigates the strengths and weaknesses of four popular HPE advanced human pose recognition libraries that can run on mobile devices: Lightweight OpenPose, PoseNet, MoveNet, and Blase Pose.

Abstract:

This article discusses the impact of global interpreter blocking on the performance of multithreaded applications in Python. The concept of subinterpreters is described as one of the solutions that allows bypassing the limitations of the GIL and ensuring effective parallel code execution. A comparative analysis of subinterpretors with traditional methods of parallel computing, such as the use of processes and threads, is carried out. The experimental results show that subinterpreters significantly increase performance under conditions of high computing loads. In addition, the article explores the use of subinterpreters in web development. The advantages of using this approach for query processing and resource management in modern web applications are considered, which can significantly improve their scalability and responsiveness. The novelty of this work lies in the in-depth analysis of subinterpreters in the context of specific use cases, which has previously not received sufficient coverage in the scientific literature. The results of the work emphasize the need for further study of subinterpreters as an alternative approach in Python, which is of interest to developers and researchers in the field of high-performance computing.

Abstract:

Amidst the rapid growth of e-commerce and increasing demand for personalization, the Russian market for customized clothing faces a shortage of technological and widely accessible solutions. This paper presents the results of a research and implementation project focused on developing a multi-brand digital platform with an integrated 3D configurator, aimed at transforming the pre-order cycle. The solution enables customers to interactively create garment designs in a web environment, while allowing designers to optimize logistics and minimize overproduction.

The primary scientific and technical contribution of this work lies in its detailed description of the platform's target architecture and a scalable 3D model processing pipeline that ensures model optimization and correct browser-based rendering. An additional contribution is the developed methodology for preparing and optimizing 3D garment models for web visualization. Formalized as a set of technical requirements, this methodology achieves a balance between visual quality and performance.

As a result of this research, the authors have addressed the challenge of unifying 3D model formats from different designers within a multi-brand digital platform—a key distinction from existing single-brand solutions. Furthermore, the implemented technology enables the customization of 3D clothing models with interactive real-time visualization of all design modifications on a single screen.

The technological feasibility and effectiveness of the solution are substantiated by a comparative analysis of existing alternatives, a market analysis using the PAM-TAM-SAM-SOM model, and an assessment of functional requirements.

The article also outlines a practical strategy for implementing the digital platform, making it a valuable resource for researchers and practitioners working at the intersection of e-commerce, computer graphics, and the digital transformation of business processes.

Abstract:

With the advent of network science, it has become possible to explore complex network systems, including social and information networks, by presenting them as graph models. The exponential growth of the total volume of scientific publications determines the relevance of the tasks of analyzing their interrelations. In network science, models and methods related to the field of so-called citation networks are being developed to solve these problems. However, network metrics are not used when analyzing publications in citation databases. The paper considers the issues of creating a decision support system for the selection of information sources based on data on the citation of scientific publications. A software package has been developed for making decisions on determining an important publication in a certain thematic area. The software package is based on a method of ranking publications by importance based on the analysis of citation networks, which allows you to identify publications that do not clearly stand out when ranking based on known bibliometric indicators or known measures of centrality of nodes in their pure form. A study and comparative analysis of software for visualization and research of all types of graphs and social networks has been conducted. Studies have been carried out confirming the effectiveness of the proposed decision support system in the selection of information sources.

Abstract:

This paper presents a method for the automated attribution of archival handwritten letters based on a Siamese neural network, addressing a key challenge in digital humanities – the authentication of historical documents. The research is motivated by the mass digitization of 17th to 19th-century archives, where attribution is often hindered by incomplete or inaccurate metadata about the authors.

The method is designed for real-world document collections and accounts for challenges typical of archival materials: poor-quality scans, significant handwriting variation, and substantial class imbalance (from 1 to over 50 samples per author). The use of a Siamese network architecture enables the extraction of discriminative vector representations (embeddings). Based on these embeddings, the method not only classifies documents by known authors but also effectively identifies manuscripts that do not match any known author in the archive. This significantly narrows down the pool of candidates for subsequent expert verification.

The study introduces a data preprocessing algorithm and provides a comparative analysis of two approaches to text analysis: at the image fragment level (300×300 px) and at the individual text line level. The developed tool offers archivists and philologists an effective solution for the preliminary sorting and attribution of handwritten documents large collections.

Abstract:

This article is devoted to the study of the possibilities of optimizing the operability and improving the efficiency of complex multifunctional software systems using the example of free automated library and information systems (hereinafter - ALIS).

By 2023, the world has accumulated valuable experience in the creation and operation of integrated ALIS of various scales and purposes, but the issues of improving their design solutions remain relevant. First of all, this concerns the need to optimize the structure of the source code in order to increase its readability and maintainability, reduce the execution time of individual functional modules, and reduce the amount of RAM used.

As part of the study, a comparative analysis of the source codes of several existing open source databases implemented in various programming languages was carried out. The main approaches to the design of the code structure were studied, the most frequently used algorithms and patterns were identified. To assess the degree of optimization of the source code, a set of indicators was developed, including an assessment of the structure, readability, modularity and other characteristics. On this basis, individual code fragments were compared before and after the use of well-known refactoring techniques.

As a result of the work carried out, it was possible to identify the most common errors and shortcomings in the structuring of the source codes of the ALIS, to determine the main directions of their optimization. Data has been obtained on the possible reduction of testing and technical support costs by improving the quality of source codes.

Abstract:

The paper assesses the limits of applicability of large language models in editorial tasks within the publishing process and identifies the optimal format of interaction between humans and algorithmic systems.

The methodological basis of the study is a comparative experiment in which several popular neural network models — Alice AI, GigaChat, DeepSeek, Gemini, and ChatGPT — performed a statistical analysis of a control text in Russian. The quantitative characteristics of the text were determined: the number of words, characters with and without spaces, and the number of paragraphs. The obtained results were compared with reference values established using the MS Word text editor, which applies a deterministic character-counting algorithm.

The results of the experiment showed that neural network models demonstrate varying degrees of accuracy when performing tasks of quantitative text analysis. The main reason for such errors lies in the architecture of large language models and the use of tokenization algorithms, which break the direct connection between characters and the model’s internal representation of the text.

Based on the results obtained, the paper proposes the concept of a hybrid architecture for publishing information systems, in which generative language models are used to perform creative and analytical tasks, while operations requiring strict formal accuracy are assigned to specialized deterministic microservices. The proposed approach makes it possible to improve the reliability and predictability of intelligent publishing systems.

Abstract:

To analyze heat fluxes, observational data for 1979-2018 were used for the North Atlantic. The spatiotemporal variability of the total heat flux was modeled by a stochastic diffusion process. The coefficients of the stochastic differential equation were estimated by using nonparametric statistics. Previously, the existence and uniqueness of a solution in the strong sense of the stochastic differential equation generated by the constructed diffusion process was proven when Kolmogorov's conditions were met. In this work, the coefficients of the equation were approximated in time by trigonometric polynomials, the amplitudes and phases of which depended on the flow values. Using a given series of 40 years in length from 1979 to 2018, spatial maps and time curves were constructed. The results are shown for 1999 and 2018, and their comparative analysis is also carried out. Numerical calculations were realized on the Lomonosov-2 supercomputer of the Lomonosov Moscow State University.

Abstract:

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.

Abstract: Since 2015, microcomputers have appeared in the information environment, which are a compact system unit with minimal functionality without peripherals. The article published the results of the analysis of the use of 6 different microcomputers in various fields of activity. The purpose of the study is to determine the limiting factors affecting the efficiency of the targeted use of microcomputers. It has been established that for scientific and educational presentations, office and trading activities, it is currently advisable to use fanless microcomputers with a perforated case and an internal WiFi antenna, at least 4 GB of operational and 64 GB of permanent memory, and a microSD (TF) memory card slot, at least 128 GB, NTFS file system), Intel HD Graphics, USB3.0 and HDMI interfaces. Based on comparative experiments, methodological recommendations were created on optimizing the configuration of the hardware-software environment of microcomputers in stationary and mobile conditions. The problems of major updates to Windows 10, as well as the compatibility of Microsoft Store software and third-party manufacturers, are analyzed. It is recommended to specialize individual microcomputers for working with 32-bit applications; accounting and cryptographic programs; as well as conducting presentations with their video. Options for optimal configuration of the Start menu of the Windows 10 desktop are suggested. It is concluded that specialization in the hardware-software configuration of modern microcomputers allows you to increase the efficiency of using single devices and their paired systems in accordance with BYOD (Bring Your Own Device).

Abstract:

This paper examines modern approaches to subword tokenization of texts as applied to the low-resource Tajik language, which is characterized by a complex morphological structure and a high degree of word-form variability. In the course of the study, a large-scale heterogeneous corpus was compiled and preprocessed, comprising 99 books and 134,497 textual articles of various genres and topics, with a total volume exceeding 33 million tokens. The corpus was cleaned of noise, normalized, and used as a basis for training and subsequent testing of subword models.

Based on this corpus, five tokenization models implementing the BPE, WordPiece, and Unigram algorithms were trained and analyzed using the Hugging Face Tokenizers and SentencePiece libraries. Comparative evaluation was conducted using a set of key metrics, including the proportion of out-of-vocabulary (OOV) words, the degree of text representation compression, tokenization speed, as well as characteristics of n-gram distribution, which make it possible to assess the ability of the models to capture the morphological and structural organization of the language. The experimental results made it possible to identify the strengths and weaknesses of different approaches to subword segmentation and to determine the most effective tokenization strategies under conditions of the morphological complexity of the Tajik language. The findings obtained can be used in the development of language models and applied NLP tools for Tajik and other low-resource languages, contributing to the expansion of their presence in the digital environment.

Abstract: The article discusses a method for implementing distributed computing using the GPU and modern administration tools and calculation management for the rendering farms. A comparative analysis is carried out between the traditional method of computing on the CPU and the modern approach to the GPU. Various configurations of the computing cluster are considered. The most productive way of rendering is determined.

Abstract:

The paper presents a comparative analysis of methods for representing three-dimensional objects to perform real-time Boolean operations in the Unity game engine environment. Four main approaches are considered: polygonal representation based on constructive solid geometry (CSG), sign distance functions (SDF), voxel methods and CAD-systems with boundary representation (B-Rep) and NURBS-surfaces.

An experimental study of the performance of polygonal algorithms of Boolean operations and SDF functions based on ray marching implementation is carried out. It is revealed that polygonal methods are characterized by high initial system construction costs, but provide stable performance during long operations and preservation of transformation results. SDF functions demonstrate high speed of operations and flexibility in creating smooth transitions between objects, but are limited in application for long-term tasks due to the peculiarities of the computational model.

The areas of effective application of each approach are identified: polygonal methods are recommended for tasks requiring precise geometric control and integration with traditional graphics pipelines, while SDF functions are optimal for procedural generation, multilayer material rendering and creation of dynamic visual effects. The results of the study can be used in the development of interactive simulators, game applications and virtual reality systems.

Abstract: The sequential nested mixed methods study focused on comparative analysis of middle school mathematics teachers’ content knowledge in two countries. The study consisted of two stages: (1) quantitative study of teacher content knowledge; (2) qualitative study of teacher topic-specific content knowledge. The initial sample for the first stage included lower secondary mathematics teachers from the U.S. (grades 6–9, N=102) and Russia (grades 5–9, N=97). The Teacher Content Knowledge Survey (TCKS) was applied to assess teacher content knowledge based on the cognitive domains of Knowing, Applying, and Reasoning, as well as addressing the lower secondary mathematics topics of Number, Algebra, Geometry, Data and Chance. The second stage – an interpretive cross-case study – aimed at the examination of the U.S. and Russian teachers’ topic-specific knowledge on the division of fractions. For the second stage, N=16 teachers (8 – from the U.S., and 8 – from Russia) were selected for the study using non-probability purposive sampling technique based on teachers’ scores on the TCKS. Teachers were interviewed on the topic of fraction division using questions addressing their content and pedagogical content knowledge. The study revealed that there are explicit similarities and differences in teachers’ content knowledge as well as its cognitive types. The study results may inform the field on priorities placed on lower secondary mathematics teachers’ knowledge in the USA and Russia. It also suggests close comparison and learning about issues related to teacher knowledge in both countries with a potential focus on re-examining practices in teacher preparation and professional development.

Abstract:

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.

Abstract:

In the modern world, the problems arising in the field of traffic are of great importance. In order to solve existing problems, various intelligent systems are being developed, one of which is the Smart City system. This work is devoted to the development of an information and analytical system (IAS) for controlling an intelligent traffic light. The presented system consists of two levels, each of which contains a set of specific operations. The first level is responsible for detecting objects, in particular pedestrians and cars at the intersection, and the second level calculates the operating time of traffic light signals for the control signal that is transmitted to the device. For comparative analysis, the combined method (HOG+SVM) Histogram of oriented gradients was chosen, based on counting the number of gradient directions on individual image areas and Support Vector Machines, which are used to construct hyperplanes in n-dimensional space in order to separate objects belonging to different classes. The results of an experimental study, during which the recognition of objects in images was carried out, showed the superiority of the developed information and analytical system over existing methods. The average accuracy of detecting pedestrians and cars through the IAS was 69.4%. In addition, according to the experiment, it was concluded that the accuracy of detecting objects in images is directly proportional to the distance from the video camera to the object.

Abstract:

Nowadays, many different tools can be used to classify images, each of which is aimed at solving a certain range of tasks. This article provides a brief overview of libraries and technologies for image classification. The architecture of a simple convolutional neural network for image classification is built. Image recognition experiments have been conducted with popular neural networks such as VGG 16 and ResNet 50. Both neural networks have shown good results. However, ResNet 50 overfitted due to the fact that the dataset contained the same type of images for training, since this neural network has more layers that allow reading the attributes of objects in the images. A comparative analysis of image recognition specially prepared for this experiment was carried out with the trained models.

Abstract:

The complexity of the problem of extracting argumentative structures is associated with such problems as selecting argumentative segments, predicting long-range connections between non-contact segments, and training on data labeled with a low degree of inter-annotator consistency. In this paper, we consider an approach to extracting argumentative relations from fairly large texts related to scientific communication. A comparative analysis was performed of fine-tuning methods using a pre-trained Longformer-type language model that takes into account long contexts and two methods that take into account annotator discrepancies in argument labeling by using the so-called soft labels obtained by uniformly smoothing labels and averaging expert assessments. The experiments were conducted on four datasets containing positive and negative examples of statement pairs (premise, conclusion) and differing in segmentation methods and average text size. The best results were obtained using the model with averaging expert assessments. At the same time, it is noted that the model using smoothed labels also increases the accuracy of classifiers, but worsens the recall.

Abstract: The use of different metaprogramming mechanisms for program libraries development in Java may have some negative effect on build time and end products. The article is aimed to describe metaprogramming mechanisms in the Java language and present the results of experiments that compare the impact of approaches on the build time of small and large projects.