Bulletin of Abai KazNPU. Series of Physical and Mathematical sciences

ABOUT THE METHODOLOGICAL AND MATHEMATICAL INSTRUCTIONS

A. Abylkassymova — 2026-03-27

This article presents a comprehensive and in-depth study of the methodological and mathematical legacy of Myrzhakyp Dulatov, a prominent educator, writer, and public figure of the early 20th century. The authors not only analyzed the publication history of the "Esep Quraly" textbooks but also explored their structural features, content, methodological principles, and pedagogical value for the Kazakh school. The study included a comparative analysis of two versions of the textbook, published between 1914 and 1928, identifying significant differences related to the construction of a concentric system, the consistent presentation of elements of arithmetic, geometry, and elementary algebra, and consideration of the specific features of the Kazakh language system, particularly in the area of numerals. The article identifies Dulatov's contribution to the development of national mathematical terminology, the expansion of opportunities for teaching mathematics in the native language, and the development of teaching methods in the context of the development of the Kazakh school. The study results showed that M. Dulatov's textbooks played a significant role in the development of mathematics education in Kazakhstan, becoming the foundation of the national methodological school, while remaining relevant for modern teaching practices focused on the cultural and linguistic characteristics of students.

METHODS FOR SOLVING TRANSCENDENTAL NON-STANDARD EQUATIONS

G.E. Berikkhanova — 2026-03-27

Formation of mathematical literacy, research, and creative competence in students is one of the main goals of mathematics education. One way to develop creative thinking skills among students is through teaching them how to solve challenging and non-standard problems, achieved through the mastery and diligence of the teacher. Non-standard transcendental problems, as well as problems solved using non-traditional methods, are often encountered in competitive and Olympiad tasks. The article considers the solution of non-standard transcendental equations using new strategic methods that go beyond traditional ones.Methods such as utilizing monotonicity, boundedness, function parity, domain of admissible values, multiplying equations by functions, and exploring numerical intervals are considered. Their application and effectiveness are demonstrated through examples. It is impossible to consider all methods for solving non-standard equations. Solving non-standard transcendental equations using the examples considered promotes students' creative understanding of the material they have learned and the development of their thinking. By using such examples in math clubs, extracurricular activities, olympiad preparation, as well as in final review sessions, students are engaged in finding effective methods for solving non-standard equations using non-traditional approaches. The work opens up new perspectives, offering a unique approach to solving equations in a modern context, and inspires students to take a creative approach to mathematics.

DEVELOPING STUDENTS’ CREATIVE THINKING IN THE PROCESS OF STUDYING PROJECTIVE GEOMETRY

Р. Жунусова — 2026-03-27

This article discusses methods for completing complex geometrical construction tasks to develop students’ creativity. The effectiveness of methods for developing creative qualities such as flexibility, lateral thinking, and a creative approach is noted. It also describes methods for developing students’ creative thinking and creative skills, problem-based learning methods, and students’ research skills. The aim of this study is to improve students’ creative abilities in lectures, practical work, and independent work on projective geometry.

In this article, we present methodological methods and approaches aimed not only at teaching but also at developing creative thinking. These methods were applied in the educational process at Margulan University in teaching projective geometry and led to a positive shift in the development of students’ creative thinking.

PEDAGOGICAL SUPPORT MECHANISMS IN TEACHING MATHEMATICS: SCAFFOLDING AS A LEARNING SUPPORT TECHNOLOGY

I. Shmigirilova — 2026-03-27

This article examines scaffolding technology as an effective pedagogical support mechanism. Schoolchildren's acquisition of mathematics is fraught with significant challenges, the need to overcome which highlights the importance of scaffold support for students and the search for didactic tools that facilitate their gradual transition to independent learning. The aim of this study is to investigate the didactic potential of scaffolding technology and the conditions for its effective integration into the educational process in the context of mathematics instruction. The theoretical study analyzed and systematized the results of scientific and methodological research on educational scaffolding, complemented by a summary of pedagogical experience using this technology in mathematics instruction. The empirical phase of the study, which included a survey of mathematics teachers, revealed the specific perceptions of scaffolding among practicing teachers, as well as the methodological and organizational difficulties associated with their lack of preparedness for the practical use of scaffold support in the educational process. The findings and recommendations presented can be used to improve the effectiveness of pedagogical support methods for students and to develop teacher professional development programs

THE NEED FOR ARTIFICIAL INTELLIGENCE-BASED TEACHING TO DEVELOP ALGORITHMIC THINKING SKILLS IN STUDENTS

B.G. Bostanov — 2026-03-27

In modern education, artificial intelligence (AI) technologies are becoming an active participant in the learning process; however, their impact on students’ cognitive abilities, particularly on algorithmic thinking, remains insufficiently studied. In this regard, the aim of the article is to present the results of a comprehensive mixed-method study aimed at the quantitative and qualitative analysis of the process of forming algorithmic thinking in students when using AI. The article also describes the application of a mixed research design, including a bibliometric analysis using the PRISMA method, a pedagogical experiment employing tests and questionnaires based on the Likert scale, as well as a qualitative thematic analysis of interviews with students. The study involved 60 students divided into control and experimental groups to examine the impact of AI on the formation of algorithmic thinking.

The results showed a statistically significant increase in the level of algorithmic thinking in the experimental group. The survey revealed positive dynamics in the development of decomposition and critical thinking skills (a shift in the mode and media). The thematic analysis of interviews demonstrated the evolution of strategies of interaction with AI–from simple queries to dialogue, verification of results, and the development of metacognitive skills. The obtained results allow us to conclude that pedagogically organized interaction with AI changes students’ thinking strategies, making them more conscious and structured, and also confirms the effectiveness of the mixed-method approach for studying this phenomenon.

POWER BI AS A DATA VISUALIZATION AND ANALYSIS TOOL IN THE DEPARTMENT'S EDUCATIONAL MANAGEMENT

G.B. Kamalova — 2026-03-27

The article is devoted to the problem of increasing the efficiency of management of the university department through the introduction of business intelligence (BI) tools. Traditional methods of data analysis are time-consuming and do not allow for prompt management decisions in a modern educational environment characterized by a large amount of data.

The purpose of this study is to substantiate the feasibility of using MS Power BI, one of the most popular business intelligence tools, to optimize the management of the university's department, develop interactive dashboards for monitoring and visualizing key performance indicators of the department (faculty publication activity, student academic performance, etc.) in real time and assess their impact on improving management efficiency. activities of the department.

The methodological basis of the research is a systematic approach, which uses a set of theoretical and experimental methods. To assess the impact of using Power BI on improving the effectiveness of department management, an experimental method of single-group preliminary and subsequent questionnaires was chosen. The sample consisted of 47 teachers of the Department of Informatics and Informatization of Education of the Kazakh National Pedagogical University named after Abai, 40% of whom have permanent general departmental duties. The survey was conducted in an online format using the GoogleForms service. The developed questionnaire included 21 questions on a 5-point Likert scale. The reliability of the results was confirmed using the Cronbach's alpha coefficient.

The results of the study. The proposed innovative method of department management based on Power BI has proven to be highly effective. It allows you to automate the generation of reports, quickly identify problem areas and trends, which, in turn, contributes to making informed management decisions. The results of the research have practical significance and will serve as a basis for implementation in the practice of department management in other educational institutions, contributing to the development of the educational management system.

ADAPTIVE EDTECH PLATFORMS FOR TRAINING TEACHERS IN IT MICRO-QUALIFICATIONS: A REVIEW OF INTERNATIONAL AND KAZAKHSTANI EXPERIENCE

Yerzat Mussirat — 2026-03-27

This article reviews the Kazakh and international experience of adaptive EdTech platforms for training teachers in IT micro-skills. In accordance with the objectives of the study, the pedagogical potential of digital platforms aimed at forming IT micro-skills in the context of continuous professional development of teachers is analyzed, and the theoretical foundations of their design and use are systematized. The following methods were used in the study: comparative analysis, content analysis and systematization. The article considers the functional capabilities, adaptability elements and design for teachers of international (Coursera, edX, Microsoft Learn, etc.) and Kazakhstani EdTech platforms. The results show that adaptive learning mechanisms play an important role in the formation of teachers along an individual learning trajectory and the development of IT micro-skills. The results of the study allow us to identify the scientific and methodological foundations of designing adaptive EdTech platforms for teachers.

ASSESSMENT OF STUDENTS’ ABILITY TO SOLVE PROGRAMMING PROBLEMS BASED ON GENAI TECHNOLOGY

Ainur Sagymbayeva — 2026-03-27

In contemporary higher education, assessing students’ programming performance requires going beyond evaluating final results to considering the cognitive and algorithmic processes involved in problem solving. The aim of this study is to develop a pedagogically grounded assessment model for evaluating students’ programming problem-solving performance using measurable indicators. The objectives of the study include identifying key problem-solving indicators in programming, linking them to assessment criteria, and experimentally validating the effectiveness of the proposed model.

The research methodology is based on a pedagogical experiment and quantitative data analysis. The experiment involved 94 undergraduate students studying Python programming. Students were assigned programming tasks differentiated by difficulty level (easy, medium, and hard), and their solutions were evaluated using seven pedagogical indicators: problem understanding, algorithmic thinking, correctness, efficiency, code quality, handling of edge cases, and explanation ability. In addition, a difficulty coefficient was introduced to account for task complexity in the final assessment.

The results indicate that as task difficulty increases, students’ correct solution rates decrease, particularly in higher-level tasks requiring efficient algorithm design, edge-case handling, and solution explanation. The findings demonstrate that the proposed assessment model enables a comprehensive evaluation of students’ programming problem-solving performance by capturing both process-oriented and outcome-based aspects. Compared to traditional assessment approaches, the model provides deeper insight into students’ strengths and weaknesses and supports more balanced and fair evaluation. The study contributes to the development of pedagogically informed assessment practices in programming education and offers a foundation for further research on adaptive and process-based evaluation models.

EMOTIONAL INTELLIGENCE AS A MEDIATOR OF THE EFFECTIVENESS OF AI-SUPPORTED PEDAGOGICAL DECISIONS IN HIGHER EDUCATION

Aliya Suguralieva — 2026-03-27

This article examines emotional intelligence (EI) as a pedagogical mediator of the effectiveness of AI-supported formative assessment practices at a university. In a sample of 64 full-time female students (one university), the following were compared over the course of a semester (T0–T2): the intensity/quality index of AI practices (personalization, early warnings, adaptive feedback, analytical dashboards), EI indicators (general latent factor and subscales: self-regulation, emotion regulation, empathy, social awareness) and educational outcomes (rubric performance, observed engagement, satisfaction/course climate). Psychometric testing included CFA and invariance tests; effects were estimated in SEM/PROCESS with bootstrapped intervals (5,000–10,000 runs) and cluster-robust errors by instructor/group. Partial mediation was observed: the indirect path through EI was statistically significant, while the positive direct effect of AI remained. The mediation rate was highest for engagement (~53%) and satisfaction (~50%), and moderate for academic performance (~36%). In the component analysis, self-regulation made the largest contribution to the indirect effect, followed by emotion regulation and empathy; the contribution of social awareness was smaller and more inconsistent. Practical conclusion: the effectiveness of AI solutions increases when integrated with EI and feedback literacy development programs (self-regulation and feedback micromodules for students; supportive digital communication for teachers), as well as when feedback attribution metrics are included in analytical dashboards. The limitations of a single institution and a gender-homogeneous sample provide direction for cross-institutional panels, longitudinal RCTs, and expansion of objective behavioral data sets, with control for covariates and robustness testing of results across model specifications.

IMPLEMENTING AI-BASED PERSONALIZED LEARNING IN MIDDLE-SCHOOL INFORMATICS: A HYBRID INSTRUCTIONAL MODEL

Niaz Tursynov — 2026-03-27

This article examines the integration of artificial intelligence–based adaptive learning systems with traditional pedagogical approaches in middle-school informatics education. The study aims to develop a hybrid instructional framework for supporting personalized learning trajectories, with particular emphasis on low-achieving students. The research is based on a systematic synthesis of selected Scopus- and Web of Science-indexed studies published between 2017 and 2025.

Using conceptual modelling and thematic analysis, the study identifies key technological, pedagogical, cognitive, and motivational factors influencing personalized learning in informatics. On this basis, a three-layer hybrid framework is proposed, integrating AI-driven diagnostics and adaptivity, teacher-led instructional mediation, and learner-centered personalized trajectories. The model also incorporates a four-stage learning cycle consisting of diagnostic assessment, pathway design, hybrid instruction, and mastery validation.

The results demonstrate that hybrid human–AI instructional models enhance learning outcomes by reducing cognitive overload, providing immediate feedback, supporting self-regulation, and strengthening learner motivation. Low-achieving students benefit from individualized task progression and sustained pedagogical support, while teachers gain access to data-informed instructional tools that facilitate targeted intervention.

The study highlights the importance of ethical governance, teacher professional development, and infrastructural readiness for successful implementation of adaptive technologies. Although the proposed framework is conceptual, it provides a theoretically grounded foundation for future empirical research and practical innovation in informatics education.

MATHEMATICAL MODELING FOR OPTIMIZING GRAIN DRYING PRIOR TO MILLING

Kh.B. Ismailov — 2026-03-27

This study presents a mathematical description of grain pre-drying formulated as a coupled system of ordinary differential equations that captures the essential heat and mass exchange between the grain and the drying agent. The model incorporates the principal operational and material parameters, including the initial moisture content of the grain and the effective heat and mass transfer coefficients. A parametric investigation is performed to quantify the influence of these factors on dehydration kinetics and on the time required to reach technologically acceptable moisture levels. The numerical predictions are evaluated against experimental measurements and demonstrate statistically acceptable agreement, supporting the adequacy of the proposed formulation for engineering analysis. The developed framework can be employed for preliminary estimation of drying duration, sensitivity assessment of process parameters, and as a foundation for subsequent extensions that account for spatial gradients of moisture and temperature in distributed drying models.

THE EQUIVALENCE PRINCIPLE AND CONSERVATION LAWS IN THE THEORY OF GRAVITY

T.B. Koshtybayev — 2026-03-27

This article presents calculations related to the covariance postulate, the equivalence principle, coordinate systems, and the conservation laws within the general theory of relativity. The free motion of a particle is analysed in the framework of space-time geometry and affine symmetry principles. The laws of free motion were evaluated using differential equations governing the relationships between the metric tensor of Euclidean space, hypersurfaces, and affine connections. The transformation of the equation of flat affine connection into the Newtonian form of the law of motion in a gravitational field is presented, along with the transformation of the equation of motion under Galilean groups. The relationships between symmetries, equations, and conservation laws of a physical system are systematized according to Noether’s theorem. Symmetry is described using Lie groups and Lie algebras, while weak conservation laws are converted into strong ones. The conservation laws in the theory of gravitation are defined by the gravitational action, and the field equation is expressed through the symmetric energy-momentum density tensor. Fundamental connections between the general and special theories of relativity, both accounting for and neglecting gravitation, are examined in the context of spatial metrics and conservation laws. The affine flat connection, absolute time, Euclidean space metric tensor, and Newtonian potential are analysed comprehensively using Cartesian coordinates on a hypersurface. All calculations are performed using the mathematical apparatus of differential geometry, tensor calculus, variational methods, and integral calculus.

ELSAKI TRANSFORMATION AND ITS APPLICATION

B. Sagindykov — 2026-03-27

The Elzaki transform is an integral transform analogous to the Laplace transform. It is a modified version of the Laplace transform and serves as a mathematical tool for analyzing engineering and physical problems. This paper explores the fundamental properties of the Elzaki transformation, which remains relatively unknown to Kazakh readers and is rarely applied in practice. Unlike conventional transforms, the Elzaki transform enables problem-solving without transitioning to a new frequency domain. In this transformation, a function of a real variable can correspond to either a real- or complex-valued function, referred to as its image. If the given function is real-valued, its Elzaki transform is generally a real-valued function as well. However, in certain cases, particularly when dealing with exponential and trigonometric functions, the image may also be complex-valued. Therefore, this study examines the key properties and theorems of the Elzaki transform to establish a set of image correspondences. As a result, we construct a table mapping original functions to their transformed counterparts. The primary objective of this paper is to demonstrate the application of the Elzaki transform in solving linear differential equations with constant and variable coefficients. Additionally, we present generalized shift theorems and analyze the transform’s effectiveness in differential equation solutions. Finally, a comparative analysis of the Laplace, Sumudu, and Elzaki transforms is conducted, highlighting their interrelationships.

STUDY OF NONEQUILIBRIUM SYSTEMS IN KINETIC THEORY USING STOCHASTIC AND FUNCTIONAL MODELS

A.M. Tatenov — 2025-03-27

This paper is devoted to the theoretical investigation of the probabilistic aspect of the kinetic theory of nonequilibrium processes in a correlated fluctuating gas. More specifically, a stochastic kinetic equation for the phase density of the system, together with the corresponding equations for its distribution and characteristic functional, as well as the equation for the generating functional, is consistently derived in a form equivalent to the irreversible kinetic equation for the distribution function. In deriving this equation, pair collisions of particles in a monoatomic gas and large-scale fluctuations in the system are considered. Within the Gibbs phase space, a multitude of dynamical trajectories in the statistical system are considered, along with non-averaged values of the single-particle state density. This spatial representation of system states reveals a functional–probabilistic formulation of classical statistical mechanics that forms the basis of the stochastic approach. The stochastic equation for the phase density is derived from the Liouville equation using a model of Markov jump processes. Although this equation is equivalent to the Boltzmann equation, it includes an additional term referred to as the nonlinear “fluctuation source.” The statistical characteristics of the resulting equation are established. The stochastic equation for the phase density of the system accounts for both dissipative and fluctuation properties and represents a generalized form of the Boltzmann–Langevin equation for unstable states of an equilibrium gas. Using hydrodynamic variables and stochastic equations, a closed hydrodynamic description of the correlation system is obtained. The relationship between the characteristic functional and the generating functional for the hierarchy of partial distribution functions is formulated through integral equations, and particular solutions corresponding to the initial conditions are presented.

PREPARATION OF DATA WITH THE HELP OF OCR FOR LLM IN KAZAKH LANGUAGE

N. Toiganbayeva — 2026-03-27

In recent years, artificial intelligence and large language models (LLMs) have undergone rapid development. The effectiveness of these models largely depends on the quality of the training data. However, the scarcity of structured text resources in the Kazakh language poses a significant challenge for LLM development. This paper explores the digitization of Kazakh-language texts using OCR technology and the creation of a high-quality dataset in JSON format. The main objective of the study is to automatically process Kazakh texts and prepare structured data suitable for training LLMs. For this purpose, scanned documents were collected, processed using Tesseract OCR, and converted into a structured JSON format. As a result, 37,062 documents were processed and used to train the LLaMA 3.2 3B model in the Kazakh language. The model demonstrated an understanding of national linguistic style and was capable of generating poetic texts. The train/loss graph indicated stable training performance.

TRANSFORMER BASED BI-LSTM DEEP LEARNING MODEL FOR AUTOMATIC CYBERBULLYING DETECTION IN KAZAKH TEXTUAL DATA

R. Abdrakhmanov — 2026-03-27

This paper presents a comprehensive study on the efficacy of a novel hybrid LSTM-CNN model for detecting cyberbullying in online social media text. The study evaluates the performance of the proposed model against traditional machine learning classifiers including SVM, Random Forest, and Decision Trees, using metrics such as accuracy, precision, recall, F-score, and AUC-ROC. The proposed hybrid model integrates the contextual processing capabilities of Long Short-Term Memory networks with the feature extraction proficiency of Convolutional Neural Networks, aiming to capture both the sequential and spatial dimensions of textual data. Results from the experiments demonstrate that the LSTM-CNN model significantly outperforms conventional classifiers, achieving high scores across all evaluation metrics. Additionally, ROC curve analyses further affirm the model's superior sensitivity and specificity in distinguishing between cyberbullying and non-cyberbullying instances. This research highlights the potential of deep learning approaches in enhancing the detection of cyberbullying, proposing a powerful tool for social media platforms to mitigate online harassment effectively. The findings also discuss the implications of deploying such advanced detection systems, considering the ethical dimensions of surveillance and privacy. Future directions include adapting the model to handle diverse linguistic contexts and exploring the integration of user feedback to refine classification accuracy. This study sets a precedent for the development of more sophisticated, context-aware technologies in the realm of digital safety and online community management.

AUTOMATION OF INFORMATION SECURITY RISK ASSESSMENT PROCESSES

S. Adilzhanova — 2026-03-27

In the context of a rapid increase in the number and complexity of cyberattacks, the need for an objective and timely assessment of information security risks is becoming increasingly critical.

The aim of this study is to develop and validate a methodology for automated risk assessment aimed at improving the accuracy, reproducibility, and efficiency of threat analysis in corporate and governmental information systems.

The methodological framework of the research combines quantitative and qualitative approaches based on international standards and models such as ISO/IEC 27005, NIST SP 800-30, and FAIR. The study employs automated monitoring and vulnerability testing systems — OpenVAS, Zabbix, Metasploit, and RiskWatch. For statistical validation of the results, the Monte Carlo method was applied within the computational environment Python 3.12 (NumPy, Pandas, SciPy).

The scientific novelty of this work lies in the development of an integrated risk assessment model that unites monitoring tools and mathematical modeling methods into a single analytical system. The practical significance of the research lies in the possibility of implementing the proposed methodology into corporate GRC and SIEM systems for continuous monitoring and adaptive risk management, as well as its applicability in educational and research activities for training specialists in cybersecurity and digital risk management.

SEMANTIC ROLE LABELING FOR KAZAKH: MODELS AND DATASETS

A.K. Aitim — 2026-03-27

A fundamental component of natural language understanding, semantic role labeling (SRL) clarifies the relationship between predicates and their arguments, therefore enabling activities including information extraction, machine translation, and question answering. Though much study has been done on SRL for high-resource languages, low-resource languages like Kazakh still relatively underexplored. This work fills the gap by offering both unique datasets and model architectures tailored specifically for Kazakh SRL. Starting with annotated SRL datasets that reflect Kazakh's rich morphological characteristics, including agglutinative suffixes and case-marking patterns, we build. Building on these data sources, we create and contrast many SRL models, from feature-driven traditional machine learning techniques to neural architectures improved by morphological embeddings. Our findings show how using Kazakh's unique language traits improves performance and draw attention to ongoing issues caused by data sparsity and complex morphology. We also address pragmatic issues for dataset generation, annotation consistency, and generalization to other Turkic languages. The findings highlight the possibility of high-quality SRL in low-resource environments and open new paths for Kazakh-language NLP study.

METHODS FOR DEVELOPING NEW MACHINE LEARNING ALGORITHMS

G.A. Alkhanova — 2026-03-27

Modern challenges in the field of data analysis require a rethinking of machine learning methods and the development of new algorithms capable of adapting to unstable and often uncertain environments. This article examines methodological aspects of designing new machine learning algorithms, including the formalization of learning tasks, model construction, loss function selection, and regularization strategies.

The need to shift from universal approaches to more context-aware architectures is substantiated, as such models can maintain a balance between accuracy, interpretability, and computational efficiency. Particular attention is given to factors limiting the applicability of existing solutions: overfitting, low noise robustness, high computational costs, and lack of decision transparency.

A classification of common challenges faced during algorithm development is proposed, divided into three levels: theoretical (modeling and justification), technical (infrastructure and resources), and applied (data quality, legal and ethical constraints).

The methodological basis of the study includes elements of systems analysis, literature review, and expert interpretation of empirical observations. The results allow for systematization of key principles in the design of machine learning algorithms and outline directions for adapting them to real-world scenarios. The findings emphasize the importance of an interdisciplinary approach that integrates mathematical methods, engineering solutions, and ethical responsibility in the deployment of intelligent systems.

COMPUTER VISION FOR STUDENT ENGAGEMENT IN ONLINE LEARNING: LITERATURE REVIEW

A.B. Bazhibayeva — 2026-03-27

This review systematizes research on automatic detection of student engagement in online learning using computer vision and deep learning from 2020 to 2025. Engagement is considered a multidimensional construct that includes emotional, cognitive, and behavioral components that play an important role in academic performance. The review analyzes the datasets used (public and specialized), feature extraction methods (facial expression, gaze direction, head posture, body movements), model architectures (CNN, LSTM, transformers), and approaches to multimodal integration. Although the transition from experimental solutions to complex real-time systems is demonstrated, persistent difficulties in model generalization, lack of diverse data, and ethical risks related to privacy and processing of personal information remain. It concludes that there is a need to develop standardized, ethically sound, and adaptable methods for assessing participation

BIBLIOMETRIC ANALYSIS OF LLM-BASED EDUCATIONAL DECISION SUPPORT SYSTEMS

A.M. Baidulla — 2026-03-27

The emergence of artificial intelligence agents as a trend is creating a need for the implementation of digital consultants based on large-scale language models (LLMs) in various fields. In the evolving educational landscape, the introduction of AI agents in this field addresses a number of pressing issues, including tailoring instruction to individual abilities, reducing teacher workload, and promoting inclusive learning. This article provides a comprehensive review of research papers from the past 15 years, drawing on a comprehensive set of leading academic databases in this field. It identifies influential authors, priority subject areas, and key research areas.

The study methodology is based on a bibliometric analysis of 1,362 articles selected from the Web of Science database for 2010-2025, designed for network visualization and cluster analysis using VOSviewer software and the R programming language.

This bibliometric analysis mapped the intellectual landscape of scientific literature on the topic of digital consultants in education based on large-scale language models. This study provides valuable insights for researchers seeking to optimize LLM-based digital assistants. The results showed that the primary focus in this area is on aspects such as improving the privatized learning and decision-making systems in educational institutions and streamlining administrative management operations. However, issues such as the long-term impact of digital assistants, scalability, data privacy, ethical issues, and the need for robust intelligent management systems still require further study.

REAL – TIME DETECTION OF CARDIAC PATHOLOGIES USING AN INTELLIGENT DIGITAL STETHOSCOPE

A.B. Baimussayeva — 2026-03-27

The paper examines the demands of the medical industry and technological advancements have combined to drive significant innovation in diagnostic tools in recent years. Presenting the creation and validation of a real-time digital stethoscope that can recognize cardiovascular issues from phonocardiograms is the aim of this study. In addition to accurately recording heart sounds, the device simultaneously analyzes acoustic characteristics to detect possible conditions. This is accomplished through the use of modern signal processing methods. Early testing on a cohort of 200 patients validated the clinical potential of this approach, showing a 94.5% diagnosis accuracy in distinguishing between pathological and normal cardiac murmurs.

This reduces the latency that comes with manual interpretation and makes it easier for people to participate in timely treatments. Given that it would make vital cardiac diagnostics accessible, it is likely that the device would be most useful in areas with limited resources. This is true even though it could be used in a variety of therapeutic settings.

The study's conclusions clearly show how important digital technology is to the integration of conventional medical equipment, opening the door to a new era of high-quality, easily accessible patient care.

NEURAL NETWORK – BASED SECURITY FRAMEWORK FOR IOT DEVICES

Gulbakhram Beissenova — 2026-03-27

The research article shows current IT advancements include the rapid growth of the Internet of Things (IoT). Internet of Things allows tangible elements to be integrated into digital frameworks. A complex infrastructure of smart sensors, sensor networks, home appliances, industrial controllers, and medical devices automates processes, improves management, and enables new services. Internet of Things use affects energy, transportation, manufacturing, healthcare, and education. The research work shows the effectiveness of a machine learning model in classifying normal and anomalous data within IoT systems.

Classification metrics, including precision, recall, and F1-score, approach 1.0, indicating high accuracy with a 100% classification rate. The model, particularly the Isolation Forest, successfully identifies anomalies in network behavior, triggering alarms when threats are detected. It highlights the advantages of using deep neural networks for malware detection in IoT environments, which can analyze both static binary files and dynamic process behaviors. These models adapt to new threats, enhancing security measures in a landscape where traditional methods fall short.

NUMERICAL SIMULATION OF FAKES SPREAD IN SOCIAL NETWORKS

Zholaman Bektemessov — 2026-03-27

The article discusses the SIHR mathematical model, which describes the process of spreading misinformation (fake) on social networks. A distinctive feature of the study is the consideration of a group of “hibernating” users who temporarily stop exchanging information but can subsequently resume activity thanks to memory mechanisms. A series of numerical experiments was conducted to assess the impact of memory parameters and failure frequency on the dynamics of the process. For various model parameters, graphs of the dependence of group density on time were constructed. The results show that the most effective way to combat misinformation is to increase the dissemination failure rate. It was also found that the processes of spontaneous and forced information recovery significantly prolong the life of a rumor on the network, and the spontaneous recovery parameter determines the final state of the system.

SECURE AUTHENTICATION SCHEMES: METHODS AND SECURITY ANALYSIS

N.A. Kapalova — 2026-03-27

This paper systematizes key methods for designing secure authentication schemes in information systems and provides a combined theoretical and practical analysis of their security properties. The study aims to define a unified threat model and to comparatively evaluate password-based, multi-factor, and passwordless authentication approaches under realistic adversary capabilities. The analysis classifies major attack vectors (online and offline guessing, credential reuse, phishing, and session takeover) and considers probabilistic models for estimating compromise success as well as compositional principles for combining factors. For password-based schemes, we show that attack success probability depends on the effective password search space, the computational cost of verifying guesses, and attempt-limiting controls. For multi-factor authentication, improved robustness is explained by a multiplicative decrease in compromise probability under factor independence. For passwordless approaches, we describe public-key challenge–response protocols that increase phishing resistance and reduce credential reuse risks. Finally, we provide practical recommendations on selecting authenticators by risk level, enforcing strict attempt limiting, enabling device binding, and applying context-aware risk assessment. The presented results can serve as a methodological basis for designing and experimentally validating strengthened authentication solutions.

APPLICABILITY OF MODERN PLANT DISEASE RECOGNITION MODELS TO FIELD CONDITIONS: A SYSTEMATIC REVIEW OF FIVE ARCHITECTURES

М.Zh. Sakypbekova — 2026-03-27

Recent advances in deep learning have significantly improved image-based methods for plant disease recognition. However, many of these systems are still developed and tested under laboratory settings, which hinders their direct applicability in real-world agriculture. This review examines 43 scientific studies published between 2020 and 2025 that applied DL methods for plant disease recognition. Unlike generalized surveys that classify articles by tasks (e.g., classification or segmentation), this work adopts an architecture-oriented approach. The methods under consideration include YOLO, Faster R-CNN, UNet, CNN+ViT hybrids, and lightweight models such as MobileNet and EfficientNet. Each architecture is evaluated in terms of structure, speed, accuracy, and performance under conditions close to field environments. One of the key objectives of this work is to identify models that are not only accurate but also practical: capable of operating in real time, on resource-constrained devices, and with images directly obtained from the field. This makes the review valuable for researchers and engineers developing AI systems for agriculture

PREDICTIVE MODELING OF EMPLOYEE BURNOUT VIA SPEECH ANALYSIS: A SYSTEMATIC LITERATURE REVIEW

Talgat Sembayev — 2026-03-27

This paper presents a systematic literature review exploring the intersection of employee burnout and speech analysis, proposing a conceptual multidimensional framework for future predictive modeling. The review adheres to the PRISMA 2020 guidelines and includes searching scientific databases, such as Scopus and PubMed, selecting and summarizing studies linking burnout, including the Maslach Burnout Inventory, and various digital phenotyping elements, such as acoustic-prosodic parameters, speech emotion recognition, and natural language processing results. Addressing the identified methodological gaps, this study outlines a theoretical framework that integrates self-supervised speech representations—specifically models like wav2vec, HuBERT, and WavLM—with emotional features, text indicators, and Organizational Network Analysis to inform future management systems. Model portability, data quality, and practical applicability are discussed separately, including cultural and linguistic specifics and personal data protection requirements in Kazakhstan (e.g., the personal data law and privacy governance approaches). The synthesized findings highlight the potential and limitations of current speech-based AI, providing a roadmap for developing ethically sound, context-aware systems for early burnout detection and preventative interventions.

MACHINE LEARNING–BASED MODEL FOR IT PROJECT COST ESTIMATION

G. Sembina — 2026-03-27

Accurate cost estimates for IT projects are critical for effective software development planning, budgeting, and risk management. Traditional cost assessment strategies, such as expert opinion, parametric models, and historical analogies, frequently fail to yield trustworthy conclusions, particularly in dynamic situations marked by rapid technological change and little historical data. This study describes the creation of a machine learning-based model intended to improve the accuracy and reliability of IT project cost estimation. The suggested approach makes use of advanced machine learning algorithms such as Random Forest and Gradient Boosting, which have been trained on datasets that include both real-world and synthetically created project data. Key project variables, including project size (LOC), team size, development duration, project complexity, and development technique, are used as predictive features. Synthetic data creation techniques are used to solve data scarcity and augment the training dataset, ensuring a greater coverage of potential project scenarios. Machine learning models outperform traditional methods in prediction accuracy, with reduced mean absolute error (MAE) and greater coefficient of determination (R²). Feature importance analysis reveals the most important aspects influencing project costs, giving vital information for project managers and decision-makers. This study advances intelligent decision support systems for IT project management and demonstrates the potential of machine learning to improve cost prediction procedures, particularly in situations when historical data is few or insufficient.

DEVELOPMENT AND IMPLEMENTATION OF AI-BASED AUTOMATED TESTING SYSTEMS FOR IMPROVING SOFTWARE QUALITY

D. Uzak — 2026-03-27

Abstract. The study proposes a hybrid AI-driven automated testing method for enhancing software quality. It employs defect prediction based on BiLSTM neural networks, test priority calculation via Q-learning, and metamorphic testing for verification in oracle-free environments. The method achieves more effective testing with greater accuracy and cost savings. System components were implemented in Python using PyTorch and OpenAI Gym, with deployment via Docker Swarm in a CI/CD environment. Experiments demonstrated a 50% reduction in testing time, 35-40% improvement in defect detection accuracy, and 60-70% decrease in human involvement compared to manual testing. The proposed approach shows practical applicability in agile environments and CI/CD pipelines.