Welcome to SDU Institutional Repository

 

Recent Submissions

Item
Image Based People Counting System for High Dense crowd
(2023) Ibrahim M.
The main objective of crowd counting is to count people in crowd images accurately. Contemporary methodologies use multi-column CNN designs to regress density maps of crowd images in order to operate the scale or perspective shifts which frequently appear in crowd images. Considering advances in technology, low-cost camera surveillance is already widely deployed. They are used for surveillance in important locations like buildings and businesses as well as in common areas like parks, schools, railways, and airports. Typically, the things in question are moving people or moving automobiles. Crowd analysis is essential in safety evaluation and other related domains. But most of the work was done by people up until this point. Automatic assessment of crowds is becoming more and more necessary when there is a lot of photographic equipment because manual approaches may become unreliable or expensive. Population counting is the first aspect of crowd evaluation that may be dealt: with as a computer vision problem. It creates persons in the image using a data image.Such detection-based systems have the drawback that the performance of the detector is strongly impacted by the presence of people in congested areas or large crowds, which reduces the accuracy of the final estimation. People have suggested clustering the trajectories of monitored visual characteristics in order to count. crowds in videos. Various columns’ reception areas respond differently to different. sized people’s bodies.our main objective is to make a system that can count high dense crowds accurately by using images of the people.
Item
Higher education students’ English language level and engagement correlation in programming courses
(2023) Mutaliyev Ye.
This thesis tries to find out if there is a link between how well students know English and how much they participate in the computer class at university. The introduction to the topic gives a general idea of how English is used in Kazakhstan. The literature study talks about things like engagement and how English is used in Kazakhstan. It also defines engagement and talks about tools like Google Forms. As part of the study method, a poll will be given to students and their answers will be analyzed. Also, grades were used in the study, and the link between English level and end course grade was looked at. The method that was used to find this connection is explained. As a result of the work, it was found that how well a student knows English has a big effect on how involved they are in the learning process in a computer class. Given how important the English language is in computing, the results give real suggestions for how to improve the level of education and the efficiency of training.
Item
Development of Recommendation System for Online Library
(2023) Makhul M.
This dissertation is devoted to the development of a recommendation system for an online library. The aim of the research is to create an efficient and personalized recommendation system that takes into account user preferences, comments in Kazakh language and likes to provide up-to-date book recommendations. The paper considers various methods of data collection, including the use of a telegram bot to generate comments in the Kazakh language and the collection of information from familiar users. Created a dataset containing 330 comments in Kazakh for model training and was divided into positive and negative comments using sentiment analysis methods. Various classification models were used for sentiment analysis, including Logistic Regression, Random Forest, Naive Bayes, and Support vector machine. The Support vector machine model achieved the highest accuracy of 95%, outperforming other models. In addition, the analysis of comments using histogram showed that positive comments usually contain more words than negative comments, which indicates more detailed and informative reviews. The identification of influential words and phrases provided insight into what aspects of books are valued by users. The developed recommendation system was integrated into the website of the online library. Two new users were created, who were given the opportunity to choose their preferred genres and languages. The system used the positive comments and likes associated with each book to generate personalized recommendations. This approach allows users to quickly find books they are interested in, which have already been popular and received positive feedback from other readers. arch has practical implications for developers of online libraries and other This rese platforms where personalized recommendations are required. The results and conclusions of this work can be used in the further development and improvement of recommendation systems, which will lead to an improvement in the quality of user service and an increase in their satisfaction.
Item
Solving the exam scheduling problems with Genetic Algorithms
(2023) Mamay K.
The exam scheduling problem is a complex task faced by educational institutions worldwide. Efficiently allocating exams within limited time slots while considering various constraints, such as student preferences, room capacities, and faculty availability, poses a significant challenge. This dissertation aims to address the exam scheduling problem by leveraging the power of Genetic Algorithms (GAs). Genetic Algorithms are robust search and optimization techniques inspired by the process of natural selection. By employing evolutionary principles, GAs have proven to be effective in finding optimal or near-optimal solutions for a wide range of combinatorial optimization problems. In this study, Ipresent a new application of genetic algorithms to solve the exam scheduling problem, to devise a solution that can be applied to Kazakhstan universities, and to evaluate the performance of GA compared to other existing algorithms commonly used in this field.
Item
Question Answering system on Regulatory Documents
(2023) Barlybay K.
The domain of legal text processing in the Kazakh language is currently underserved, presenting a unique challenge due to its specialized language and the relative scarcity of computational resources dedicated to it. This thesis explicitly identifies the problem: the need for an efficient model to process, understand, and generate meaningful insights from Kazakh legal texts. Addressing this problem, the thesis proposes a solution by developing and evaluating bespoke language models pre-trained on a vast corpus of Kazakh legal documents. The study begins with the assembly of a corpus, which comprises over 315 million words from Kazakh legal texts, alongside a benchmark dataset of 2500 multiple-choice questions for civil service examinations in Kazakhstan. Three language models based on the BERT architecture are then pre-trained. Among these, one model is pre-trained entirely from scratch. To emulate a real-world application in the legal domain, the performance of these models is assessed using the multiple-choice question-answering task. The BERT base model pre-trained from scratch, leveraging both Masked Language Modeling (MLM) and Next Sentence Prediction (NSP) tasks, achieves an accuracy of 56.11%. This result underlines the potential of custom pre training strategies on domain-specific corpora for enhancing the performance of language models in specialized areas. In conclusion, this research represents a significant advancement in using AI for legal text processing in the Kazakh language. It presents a promising solution to the problem, paving the way for more efficient and informed decision-making processes in legal and civil service settings.