CENTRAL ASIAN JOURNAL OF MATHEMATICAL THEORY AND COMPUTER SCIENCES

Design and Development of a Python-Based Multi-Game Desktop Application with CustomTkinter

2026-08-02T09:33:22+00:00

The current work is a Graphical User Interface (GUI) based Game Center application developed using Python and CustomTkinter framework. The program includes a user authentication system (login and signup) that is connected to a MySQL database using the `pymysql` module to securely manage user credentials. After the user has been successfully authenticated, he/she can access the centralised game hub which consists of three interactive games, namely, Hand Cricket, Rock Paper Scissors and Tic Tac Toe. Separate logic and user interfaces that are easy to understand are built for each game with `CustomTkinter` which increases the visual appeal and interactivity. The Hand Cricket game is a simplified cricket match game in which the user plays against the computer with the help of random number generation. Rock Paper Scissors generates dynamic action through images and randomisation . Tic Tac Toe offers a grid - based challenge for two players with automatic win and draw detection . On the whole this project is a good example of how GUI and DB management, alongside basic game logic, can be combined into an application. It prepares the ground for building multi-game platforms and shows the potential of Python for building fun, visually stunning desktop apps.

The Effect of Metacognitive Strategies on Achievement and the Development of Metacognitive Skills in Sixth-Grade Female Students in Mathematics

2026-08-03T12:30:52+00:00

The current study aimed to explore the impact of metacognitive strategies on achievement and the development of metacognitive skills among sixth-grade elementary students in mathematics. To achieve this goal, the researcher used a quasi-experimental method with partial control by selecting two equivalent groups: one experimental group taught using metacognitive strategies and one control group taught using the regular method, with 27 students in each group. The research tools consisted of an achievement test with 30 multiple-choice items covering the prescribed math topics, and a metacognitive skills test composed of 25 items, the two tests were subjected to validity procedures by being presented to a group of specialized experts. The researcher applied the two tools on a pilot sample to ensure the clarity of the items and to extract the difficulty and discrimination indices, as well as the effectiveness of the wrong alternatives for each item. She also extracted the reliability of the two tests using the Kuder-Richardson KR-20 formula. To test the research hypotheses, the researcher used the t-test for two independent samples. The results showed that the students in the experimental group outperformed in both the achievement test and the metacognition test, indicating the effectiveness of metacognitive strategies in improving academic achievement and developing metacognitive skills among sixth-grade students in mathematics, which supports improving math teaching methods and enhancing learning outcomes.

Comparative Analysis of Contingency Models in Medical Data using Log-linear Models and Logistic Regression Models

2026-08-07T11:30:11+00:00

This research explored the use of linear logistic studies and binary and multi-logarithmic reference models in analyzing n×m Tables Contingency for the development of relationships between genders. It also examined how each model interprets the relationships between these correlations. This research has applications in the medical field, where the effects of age and gender on weight were studied, and their impact on weight prediction was investigated. Logarithmic models identified weight gain in any age group, while logistic models developed a predictive equation for the effect of age and gender on weight. Each model has its own characteristics logarithmic models focus on studying the relationship between variables, while logistic models construct a predictive equation to interpret the results.

Facial Emotion Recognition using Convolutional Neural Networks

2026-08-09T12:57:17+00:00

Facial Emotion Recognition (FER) is an important and difficult topic of computer vision and artificial intelligence. This project employs a deep learning algorithm, Convolutional Neural Network (CNN), for the classification of human facial expressions. The model is trained on the “Facial Expression Dataset” hosted on Kaggle. The dataset contains 48x48 greyscale images of faces classified into seven different emotions: happy, sad, angry, fear, disgust, neutral and surprise. Load dataset and organise paths and labels into a Pandas DataFrame Performing EDA (Exploratory Data Analysis) to visualise the class distribution and sample images. Our robust pre-processing pipeline loads images as grey-scale, normalises the pixel values to the range [0, 1], and one-hot encodes the categorical labels. Designing a sequential CNN architecture that includes three convolutional blocks (Conv2D, MaxPooling2D, Dropout) and multiple fully connected (Dense) layers. The model is trained with Adam optimiser, categorical_crossentropy loss and Early Stopping to prevent overfitting. The trained model demonstrates the ability to learn and classify facial expression well. Finally, the project saves the trained model (facial_emotion_model.keras) for future inference and application.

Offline Learning with Smart Technology: An Inclusive Digital Education Platform for Rural Students

2026-08-12T06:10:33+00:00

Offline Learning with Smart Technology is an innovative educational app designed to facilitate accessible and inclusive learning for students in rural and remote areas with limited or no internet connectivity. The platform offers an interactive learning environment with lessons, quizzes, multimedia educational content, and features to track progress, all available offline. It aims to foster digital literacy as well as critical thinking, problem-solving skills and independent learning among students in need of online educational resources. Its intuitive user interface enables learners to easily access educational materials and participate in structured learning activities without the need for continuous internet access. Interactive quizzes provide students immediate feedback allowing them to identify their strengths and weaknesses. Progress reports can monitor the learning development over time. Teachers can also get information about student performance and progress thru the platform, which helps them provide better guidance and academic support. The proposed system combines smart technology with offline access to bridge the digital education gap between urban and rural communities. To summarise, the application presents a practical technology-driven solution for inclusive education, enhancing access to digital learning resources and enabling better learning outcomes in underserved communities.

Lecture Buddy: An AI-Powered Intelligent Assistant for Automated Study Material Generation

2026-08-14T05:35:08+00:00

Students sometimes find it difficult to organise and make efficient use of their lecture notes in the fast-paced academic environment of today. Traditional note-taking is time-consuming and can lead to scattered and unstructured material which is hard to revise properly. Lecture Buddy is an AI-powered educational aide that addresses these issues by translating raw lecture notes into organised, interactive study resources. The system enables students to upload their lecture notes and then intelligently generates succinct summaries, study flashcards and multiple-choice tests. It uses OpenAI’s language models and cutting-edge natural language processing techniques to break down complicated ideas and organise them in a way that is easy to grasp and remember. Summaries distilll lengthy notes into essentials. Flashcards facilitate active memory. Quizzes allow students to check their understanding. The outcome is a more interesting and efficient learning experience. Also, students of different courses can use Lecture Buddy making it a flexible tool for exam preparation and everyday study. The main aims of the software are to save time; boost understanding and develop good study habits. Combining AI-driven content generation with common study habits, Lecture Buddy helps students enhance their productivity and improve their academic results. The software should be improved in the future to allow saving notes by subject, automatic organization and customised educational recommendations.

AI-Powered Intelligent Expense Tracking and Predictive Financial Management System

2026-08-15T08:26:51+00:00

AI Based Expense Tracker App is a smart financial management software that automates expense tracking and provides data-driven information to users. Using advanced artificial intelligence and machine learning models, the app simplifies personal and business finance management by automatically categorizing transactions, extracting data from receipts, and predicting future spending patterns. The app leverages Optical Character Recognition (OCR) and Natural Language Processing (NLP) to scan and interpret bills, invoices, and digital payment data. It integrates securely with multiple bank accounts, credit cards, and e-wallets, ensuring real-time expense synchronization. Users receive predictive budget alerts, spending summaries, and personalized insights to help manage and reduce unnecessary expenditures. AI models like BERT for text classification and LSTM neural networks for spend forecasting achieve over 94% categorization accuracy and a predictive variance of just ±6%. Visual dashboards and trend analytics empower users to make informed financial decisions while maintaining transparency and control over their money flow. The system emphasizes data security and privacy compliance, aligning with GDPR and RBI guidelines. Continuous learning mechanisms allow the AI to adapt to user behavior, improving accuracy and personalization over time. Future developments include an AI voice assistant, investment recommendations, and team-level financial dashboards for businesses. In essence, the AI-Based Expense Tracker App combines automation, predictive intelligence, and user-centric design to transform the way individuals and organizations monitor and manage expenses—promoting financial discipline and smarter decision-making through technology-driven insights.

AI-Powered Automated Question Paper Generation System Using OpenAI and Web Technologies

2026-08-20T08:27:01+00:00

The Smart Automated Assessment Paper Generator is an AI-based system that makes the process of generating question papers for educational institutions easier and more modern. Traditionally, generating question papers requires significant manual effort from educators to ensure that the questions cover the topic, are balanced in difficulty and use diverse question formats. This project eliminates these challenges by automating them using Artificial Intelligence and modern web technologies. The system is built with Node.js for the back end, React.js for the user interface, and the OpenAI API for generating questions. The user can enter the subject, topic, type of question, level of difficulty and time they want. Based on these parameters, it automatically generates a well-structured question paper comprising diverse, contextually apt questions. OpenAI’s sophisticated language model allows for relevant, coherent and logically structured questions that adapt to the selected level of complexity. The system is designed to be both functional and usable, with an intuitive interface that enables users to conveniently customize, view and export the generated paper. It reduces effort and time in paper-based settings, eliminates duplication and ensures quality consistency across multiple assessments.

The Co-e-filters among C^* (X)

2026-08-19T21:33:35+00:00

Here we investigate the Co-ϵ-filter sometimes we say Co-e-filter (filters of Cozero e-set) and e-ideal association of C^* (X) for real-valued bounded continuous functions on Tychonoff space X, the essential concept of the study is Coz-filters of bounded continuous functions herein denoted by Co-e-filters for non-compact spaces. We establish the relationship among Co-ϵ-ultrafilters and the minimal prime ideals of C^* (X) providing a direct analogue to the known relationship between e-ultrafilters and maximal ideals, the central result demonstrates a precise correspondence between these co-e-ultrafilters and the minimal prime z-e-ideals of C^* (X), thereby mirroring the established association in C(X) within the bounded context.

Preserving Monotonicity of Two-Dimensional Data Using a Three-Parameter Rational Cubic Spline

2026-08-25T04:31:14+00:00

Preserving the original form of the data whilst inserting new data is a fundamental challenge in many applications, particularly as a number of common insertion methods used in practical applications result in distortions in the graphical representation of the data. This research proposes a fractional spline function as an interpolation function designed to ensure the smoothness of two-dimensional data whilst avoiding certain physical artefacts such as oscillation, as well as providing the ability to control the shape of sub-intervals locally. The proposed function features three positive parameters that can be adjusted manually, thereby allowing control over the tension or curvature of the graphical curve as required. The method relies on the use of the first derivative in its mathematical sense to ensure a smooth transition between each sub-interval of the function’s curve, resulting in a final shape that is smooth and free from physical artefacts. The research includes numerical examples and comparisons demonstrating that the proposed method offers high efficiency in maintaining the regularity of the function, whilst also providing flexibility, reliability and computational efficiency.

On Weak Γ Multiplication and Weak Γ Comultiplication SΓ-Acts Over Commutative Γ-Semigroups

2026-08-27T14:04:48+00:00

This paper introduces and investigates weak Γ multiplication SГ-acts and weak Γ comultiplication SГ-acts over commutative Γ semigroups, notions that generalize classical weak multiplication and comultiplication modules. We establish fundamental properties, provide equivalent characterizations using annihilators and residual Γ ideals, and prove a localization theorem. It is shown that every finitely generated weak Γ multiplication SГ-act is cyclic, hence a multiplication act. Over Γ-artinian Γ semigroups, every weak Γ multiplication SГ-act is cyclic. A categorical duality between finitely generated faithful weak Γ comultiplication SГ-acts and finitely generated faithful multiplication SГ-acts is established for semihereditary Γ semigroups. Several examples delimit the concepts introduced.

Algebraic Structures in Fractional Differential Equations for Concentration-Dependent Transport

2026-08-23T09:52:10+00:00

This research explores the role of algebraic structures in analyzing and solving partial differential equations for concentration-dependent transportation models. The introduction defines the research problem while highlighting the importance of algebraic structures in analyzing and understanding partial differential equations for concentration-dependent transportation models and provides a simplified explanation of the partial differential equations and their applications in the fields of transportation focus-based. The theoretical framework section consists of two main parts. First, an introduction to partial differential equations and their importance is provided, along with their applications in concentration-dependent transport models. Second, the theoretical basis for solving differential equations is discussed and the mathematical tools necessary for solving partial differential equations and their application are reviewed. The methodology is divided into sections of theoretical analysis and case study. The theoretical analysis reviews the theories and algebraic methods used in analyzing partial differential equations for concentration-dependent transport models. The case study applies algebraic structures to a specific partial differential equation for a concentration-dependent transport model, along with analyzing the results. The Results and Discussion section provides an analysis of the theoretical and experimental results obtained from using algebraic structures in solving partial differential equations. The Discussion section explores the effect of using algebraic structures on the accuracy of the results and practical applications. Finally, the Conclusions and Recommendations section summarizes the main findings and conclusions and makes recommendations for future research and practical applications. Overall, this research paper delves into the application of algebraic structures in solving partial differential equations for concentration-dependent transport models with an emphasis on their role in enhancing the accuracy of results and practical applications in this field.

Proposed Hybrid Intelligent Model Based on Artificial Neural Networks and Decision Trees for Early Detection of Learning Disabilities: Design and Performance

2026-09-02T07:37:43+00:00

Early identification of learning difficulties is an important challenge in education because timely screening can support appropriate educational intervention. Machine-learning methods can process multidimensional information, but individual approaches involve trade-offs between predictive performance and interpretability. This study proposes a sequential hybrid model that combines a C4.5 decision tree with a multilayer perceptron (MLP). In the first stage, the decision tree identifies informative features and generates human-readable IF–THEN rules; in the second stage, the selected features are provided to an MLP for three-class classification. The model was evaluated on a synthetic dataset of 1,200 children aged 5–7 years, containing 18 simulated demographic, cognitive, linguistic/academic, and behavioral features and three target classes: Typical, At-Risk Dyslexia, and At-Risk Dyscalculia. The hybrid model achieved 96.8% accuracy and a macro F1-score of 0.97 on the reported evaluation set, compared with 94.2% and 0.94 for the standalone MLP and 88.6% and 0.88 for the standalone decision tree. Because the experiments used synthetic data exclusively, these findings constitute a methodological proof of concept rather than evidence of clinical or educational diagnostic validity. Validation using independently collected real-world datasets is required before practical deployment.

Discretization and Adaptive Control of a Variable-Order Fractional Burgers Equation with Short-Memory Acceleration

2026-09-05T12:55:17+00:00

We study a fractional generalization of Burgers’ equation where the time derivative is replaced by a variable-order Grünwald–Letnikov (GL) fractional derivative . This naturally introduces a memory effect that depends on both space and time. We develop a consistent numerical scheme combining a discretization of the variable-order GL operator with a semi-implicit upwind finite-difference method for the nonlinear convection term, solved efficiently via a tridiagonal Thomas algorithm. Beyond the baseline scheme, we explore three practical extensions: a short-memory truncation that reduces runtime by an order of magnitude with minimal accuracy loss (≲1% deviation); a state-dependent adaptive order that sharpens fronts automatically; and a careful discussion of the inherent ambiguity in defining variable-order GL derivatives. Numerical experiments validate the method and demonstrate its flexibility across a range of test cases. ‎

Smart Commute: An AI-Powered Real-Time Transit Tracking, Route Optimization, and Safety System

2026-09-07T11:01:35+00:00

The paper is a cutting-edge Web-based application aimed at transforming urban transportation through real-time GPS tracking, AI-based route optimisation, and crowd density reporting. The system solves important urban travelling problems like delay in reaching destination, overcrowded vehicle, safety issues and route planning. The platform leverages Java (backend services), Python (data processing & machine learning) and MySQL (database) to provide dynamic, data-driven solutions for commuting. Smart Commute uses live transit data and smart algorithms to provide users with the fastest, safest and least-crowded ways to get around. Plus, the app has safety features such as route alerts and emergency contact notifications, which help make the overall travel experience better. This project is an illustration of the extent to which modern programming methodologies and intelligent data utilisation can be used to transform traditional transport systems into intelligent, commuter friendly ecosystems.

Model and Measure the Volatility of the Iraqi Dinar’s Exchange Rate Against the US Dollar Using the AR(1)-GARCH(1,1) Model Under the Student’s t-Distribution for the Period 2015–2024

2026-09-08T16:51:33+00:00

This study aims to model and measure the volatility of the Iraqi dinar’s exchange rate against the US dollar for the period 2015–2024, using modern standard methods for financial time series. The research problem stems from the inadequacy of traditional linear models, such as Ordinary Least Squares (OLS), and the assumption of a normal distribution in capturing the extreme shocks experienced by the Iraqi economy. To overcome this limitation, the study employed an autoregressive model with conditional heteroscedasticity (AR(1)-GARCH(1,1)) model, assuming a Student’s t-distribution. The results of the statistical tests showed a decisive rejection of the normality hypothesis based on the Jarque–Bera test, and provided mathematical evidence of the presence of a clustering phenomenon in volatility based on the ARCH-LM test. The estimation results also confirmed that the probability distribution of the dinar exchange rate returns is characterised by heavy tails with a degree of freedom of (ν=5.24), implying a high probability of severe shocks. The study concludes that conditional variance models are the most effective in measuring the exchange rate risk of the Iraqi dinar.

Python-Based Web Scraping Tool for Real-Time Weather and News Information Extraction and Analysis

2026-09-09T13:40:25+00:00

This project presents a Python-based web scraping tool that can automatically extract real-time weather information and news headlines from various internet sources. This tool makes that easy by bringing all the info together in one place, so you don’t have to go looking. The system can efficiently collect and organise data using Python libraries like requests, Beautiful Soup and pandas. The requests library is used to get the content of web pages and Beautiful Soup is used to parse and extract useful information like temperature, humidity, wind speed and weather conditions from weather websites and top headlines and brief descriptions from popular news portals. The scraped data is cleaned, structured and presented in a user-friendly format which is easy to interpret and analyse. This guarantees users accurate, nicely formatted results that can be stored or further processed for other applications. To summarise, this project demonstrates the usefulness of web scraping as an effective way to automate data collection and deliver real-time information in a timely fashion. The weather and news extraction tool is a perfect solution for everyday information retrieval, taking advantage of the simplicity of Python and the extensive library support. It reduces manual work, makes things more affordable and serves as the basis for creating more sophisticated systems that leverage real-time data to analyse, predict and decide.

Radical-Constrained Intersections of Direct Summands in Modules

2026-09-10T07:49:49+00:00

We introduce the radical-constrained intersection property (RCI), a radical-localized weakening of the summand intersection property. A right R-module M is called RCI when, for every pair of direct summands A,B ≤⊕ M satisfying A ∩ B ≤ Rad(M), the intersection A ∩ B is a direct summand of Rad(M). This formulation localizes the splitting requirement itself inside the Jacobson radical and avoids lifting-type approximations by a possibly zero direct summand. We prove that every SIP-module is RCI, that the implication is strict, and that modules with semisimple or zero radical are RCI. The class is closed under direct summands, but not under arbitrary finite direct sums; a sufficient closure condition is obtained for Hom-orthogonal direct sums. We also give a quotient theorem under an explicit direct-summand lifting hypothesis and an idempotent characterization on the radical. The examples Z₂ ⊕ Z₄ and Z₄ ⊕ Z₈ show, respectively, that RCI need not imply SIP and that finitely generated modules need not be RCI. Relations with SISZ, SIPr, SIP-lifting, and radical-supplemented modules are discussed to clarify the position of the new condition in current decomposition theory.