Soil erosion poses a serious environmental and agricultural threat that undermines land productivity, sustainability, and ecosystem stability. This study develops a robust machine learning framework for predicting and analyzing soil erosion across diverse landscapes by integrating advanced remote sensing data, climate indicators, and soil characteristics. Spectral indices such as the normalized difference vegetation index (NDVI), moisture stress index (MSI), and surface albedo were employed to assess vegetation condition, moisture levels, and surface reflectance. The proposed model, based on the extreme gradient boosting (XGBoost) algorithm, classifies erosion stages with up to 99% accuracy, ranging from healthy land to severely degraded areas. The methodology includes comprehensive feature engineering, dataset preprocessing, and model evaluation. Furthermore, a comparative analysis with traditional models (USLE and RUSLE) highlights the superior predictive performance of the proposed approach. The findings offer valuable insights for sensor-based monitoring systems and cloud-based decision-support tools, supporting sustainable land use management, erosion risk mitigation, and effective soil conservation strategies.

This study explores how university students experience learning programming through educational games in extracurricular computer science education. A qualitative phenomenological design was employed to interpret the meanings students attribute to game-based learning rather than to evaluate instructional effectiveness. Data were collected through semi-structured interviews with 60 undergraduate students who participated in supervised extracurricular programming activities involving educational coding games. The analysis revealed that educational games transform programming from a formal academic subject into an exploratory problem-solving activity supported by immediate feedback. Students described a reduction of anxiety and perceived the environment as psychologically safe for experimentation. However, gameplay alone produced an intuitive yet fragmented understanding, requiring instructor explanation for conceptual structuring. The experience of learning was therefore characterized as a transition from experiential discovery to reflective comprehension. The findings indicate that educational games function as an experiential entry point through which students begin to interpret programming as a meaningful activity, while instruction supports conceptual articulation. This study contributes to understanding the pedagogical role of educational games not as substitutes for teaching but as mediating environments shaping students' perception of programming.

The Working Set concept, originally introduced by P. Denning for memory management, defines a dynamic subset of system elements actively in use. Designed to reduce page faults and prevent thrashing, it has proven effective in optimizing memory performance. This study explores the interdisciplinary potential of the Working Set by applying it to two distinct domains: virtual memory systems and epidemiological modeling. We demonstrate that focusing on the active subset of a system enables optimization in both contexts—minimizing page faults and containing epidemics via dynamic isolation. The effectiveness of this approach is validated through memory access simulations and agent-based epidemic modeling. The advantages of using the Working Set as a general framework for describing the behavior of dynamic systems are discussed, along with its applicability across a wide range of scientific and engineering problems.

This study proposes a machine learning-based approach for automatic summarization of scientific documents using a fine-tuned DistilBART model a lightweight and efficient version of the bidirectional and auto-regressive transformers (BART) architecture. The model was trained on a large corpus of 12,540 scientific articles (2015–2023) collected from the arXiv repository, enabling it to effectively capture domain-specific terminology and structural patterns. The proposed pipeline integrates advanced text preprocessing techniques, including tokenization, stopword removal, and stemming, to enhance the quality of semantic representation. Experimental evaluation demonstrates that the fine-tuned DistilBART achieves high summarization performance, with ROUGE-2=0.472 and ROUGE-L=0.602, outperforming baseline transformer-based models. Unlike conventional approaches, the method shows strong applicability beyond academic research, including automated indexing of technical documentation, metadata extraction in digital libraries, and real-time text processing in embedded natural language processing (NLP) systems. The results highlight the potential of transformer-based summarization to accelerate scientific knowledge discovery and improve the efficiency of information retrieval across various domains.

This study aims to develop a hybrid system for the automatic annotation of scientific texts that efficiently processes multilingual publications using state-of-the-art natural language processing (NLP) technologies. The system integrates classical algorithms (Gensim, NLTK) with transformer-based models via the Cohere API to achieve high semantic consistency and accuracy in annotations. The system architecture comprises modules for data acquisition, preprocessing, manual and automatic annotation, data storage, and quality control. The performance of the proposed model was benchmarked against established methods such as BERTSUM, TF-IDF + LSA, and GPT-3.5-turbo using evaluation metrics including ROUGE, BLEU, and METEOR. The hybrid model outperformed other automated systems, demonstrating superior scores across ROUGE-1 (0.52), BLEU (0.41), and METEOR (0.39) metrics, indicating its effectiveness in producing concise and semantically accurate summaries. The system also achieved 100% language detection accuracy and 90% accuracy in semantic word relationships via Word2Vec. The integration of traditional statistical methods with advanced transformer models enables the proposed system to deliver high-quality annotations suitable for diverse scientific domains. The results validate the model’s ability to process and summarize complex scientific texts effectively. This system provides a scalable, secure, and user-friendly platform for researchers, institutions, and developers. It supports multilingual annotation, seamless API integration, and potential deployment in cloud environments, offering significant benefits for academic, biomedical, and information-intensive sectors.

This article addresses the development of a biometric personal identification system based on palm parameter analysis using machine learning technologies. The relevance of the study is determined by the need to ensure information security, access control, and reliable user identification. Since the palm as a biometric identifier possesses a high degree of individuality, its anatomical features–such as finger contours, skin texture, palm lines, and blood vessel patterns–are used as effective characteristics for personal recognition. In the course of the study, the ROI-LANet model was employed to automatically identify significant biometric regions in palm images. This model enables accurate localization of key palm points, image alignment, and extraction of regions of interest (ROI). User classification based on the extracted ROI was performed using the deep convolutional neural network ResNet18. The residual block-based architecture allows efficient recognition of complex biometric features and improves classification accuracy. Experimental results demonstrated that the proposed system achieves high accuracy and stability. During testing, the identification accuracy reached approximately 97%. In addition, the system showed robustness to variations in lighting conditions and image acquisition angles. The obtained results indicate a high potential for practical application of the developed biometric system in security systems, access control, and transaction verification.