Arailym Tleubayeva

Ms. Arailym Tleubayeva

School of Artificial Intelligence and Data Science, Astana IT University, Nur-Sultan, Astana, Kazakhstan
orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 June 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
SBSvitlana Biloshchytska
Svitlana Biloshchytska
Arailym TleubayevaArailym Tleubayeva
Arailym Tleubayeva
Oleksandr KuchanskyiOleksandr Kuchanskyi
Oleksandr Kuchanskyi
ABAndrii Biloshchytskyi
Andrii Biloshchytskyi
Yurii AndrashkoYurii Andrashko
Yurii Andrashko
Saltanat SharipovaSaltanat Sharipova
Saltanat Sharipova
This study presents an advanced hybrid approach for detecting near-duplicate texts in the Kazakh language, addressing the specific challenges posed by its agglutinative morphology. The proposed method combines statistical and semantic techniques, including N-gram analysis, TF-IDF, LSH, LSA, and LDA, and is benchmarked against the bert-base-multilingual-cased model. Experiments were conducted on the purpose-built Arailym-aitu/KazakhTextDuplicates corpus, which contains over 25,000 manually modified text fragments using typical techniques, such as paraphrasing, word order changes, synonym substitution, and morphological transformations. The results show that the hybrid model achieves a precision of 1.00, a recall of 0.73, and an F1-score of 0.84, significantly outperforming traditional N-gram and TF-IDF approaches and demonstrating comparable accuracy to the BERT model while requiring substantially lower computational resources. The hybrid model proved highly effective in detecting various types of near-duplicate texts, including paraphrased and structurally modified content, making it suitable for practical applications in academic integrity verification, plagiarism detection, and intelligent text analysis. Moreover, this study highlights the potential of lightweight hybrid architectures as a practical alternative to large transformer-based models, particularly for languages with limited annotated corpora and linguistic resources. It lays the foundation for future research in cross-lingual duplicate detection and deep model adaptation for the Kazakh language.

30 October 2025 in Information
Verified
Journal Article
Open access
PDF file
AMAigerim Mansurova
Aigerim Mansurova
Arailym TleubayevaArailym Tleubayeva
Arailym Tleubayeva
ANAliya Nugumanova
Aliya Nugumanova
ASAday Shomanov
Aday Shomanov
SSSadi Evren Seker
Sadi Evren Seker
This paper presents a systematic evaluation of large language models (LLMs) and retrieval-augmented generation (RAG) approaches for question answering (QA) in the low-resource Kazakh language. We assess the performance of existing proprietary (GPT-4o, Gemini 2.5-flash) and open-source Kazakh-oriented models (KazLLM-8B, Sherkala-8B, Irbis-7B) across closed-book and RAG settings. Within a three-stage evaluation framework we benchmark retriever quality, examine LLM abilities such as knowledge-gap detection, external truth integration and context grounding, and measures gains from realistic end-to-end RAG pipelines. Our results show a clear pattern: proprietary models lead in closed-book QA, but RAG narrows the gap substantially. Under the Ideal RAG setting, KazLLM-8B improves from its closed-book baseline of 0.427 to reach answer correctness of 0.867, closely matching GPT-4o’s score of 0.869. In the end-to-end RAG setup, KazLLM-8B paired with Snowflake retriever achieved answer correctness up to 0.754, surpassing GPT-4o’s best score of 0.632. Despite improvements, RAG outcomes show an inconsistency: high retrieval metrics do not guarantee high QA system accuracy. The findings highlight the importance of retrievers and context grounding strategies in enabling open-source Kazakh models to deliver competitive QA performance in a low-resource setting.

30 September 2024 in Scientific Journal of Astana IT University
Verified
Journal Article
Open access
PDF file
Arailym TleubayevaArailym Tleubayeva
Arailym Tleubayeva
ASAday Shomanov
Aday Shomanov
This paper presents a comparative analysis of large pretrained multilingual models for question-answering (QA) systems, with a specific focus on their adaptation to the Kazakh language. The study evaluates models including mBERT, XLM-R, mT5, AYA, and GPT, which were tested on QA tasks using the Kazakh sKQuAD dataset. To enhance model performance, fine-tuning strategies such as adapter modules, data augmentation techniques (back-translation, paraphrasing), and hyperparameter optimization were applied. Specific adjustments to learning rates, batch sizes, and training epochs were made to boost accuracy and stability. Among the models tested, mT5 achieved the highest F1 score of 75.72%, showcasing robust generalization across diverse QA tasks. GPT-4-turbo closely followed with an F1 score of 73.33%, effectively managing complex Kazakh QA scenarios. In contrast, native Kazakh models like Kaz-RoBERTa showed improvements through fine-tuning but continued to lag behind larger multilingual models, underlining the need for additional Kazakh-specific training data and further architectural enhancements. Kazakh’s agglutinative morphology and the scarcity of high-quality training data present significant challenges for model adaptation. Adapter modules helped mitigate computational costs, allowing efficient fine-tuning in resource-constrained environments without significant performance loss. Data augmentation techniques, such as back-translation and paraphrasing, were instrumental in enriching the dataset, thereby enhancing model adaptability and robustness. This study underscores the importance of advanced fine-tuning and data augmentation strategies for QA systems tailored to low-resource languages like Kazakh. By addressing these challenges, this research aims to make AI technologies more inclusive and accessible, offering practical insights for improving natural language processing (NLP) capabilities in underrepresented languages. Ultimately, these findings contribute to bridging the gap between high-resource and low-resource language models, fostering a more equitable distribution of AI solutions across diverse linguistic contexts.

1 October 2024 in International Journal of Electrical and Computer Engineering (IJECE)
Verified
Journal Article
Open access
PDF file
AOAinur Orazayeva
Ainur Orazayeva
JTJamalbek Tussupov
Jamalbek Tussupov
GSGulmira Shangytbayeva
Gulmira Shangytbayeva
AGAssem Galymova
Assem Galymova
UZUlzhalgas Muratbekovna Zhunissova
Ulzhalgas Muratbekovna Zhunissova
ZKZhanat Kenzhebayeva
Zhanat Kenzhebayeva
This work is devoted to the research and development of methods for effectively identifying breast pathologies using modern machine learning technologies, such as you only look once (YOLOv8) and faster region-based convolutional neural network (R-CNN). The paper presents an analysis of existing approaches to the diagnosis of breast diseases and an assessment of their effectiveness. YOLOv8 and Faster R-CNN architectures are then applied to create pathology detection models in mammography images. The work analyzed and classified identified breast pathologies at six levels, taking into account different degrees of severity and characteristics of the diseases. This approach allows for more accurate determination of disease progression and provides additional data for more individualized treatment planning. Classification results at various levels can improve the quality of medical decisions and provide more accurate information to doctors, which in turn improves the overall efficiency of diagnosis and treatment of breast diseases. Experimental results demonstrate high accuracy and speed of image processing, providing fast and reliable detection of potential breast pathologies. The data obtained confirm the effectiveness of the use of machine learning algorithms in the field of medical diagnostics, providing prospects for the further development of automated systems for detecting breast diseases in order to improve early diagnosis and treatment efficiency.

4 May 2023
Verified
Conference Paper
Closed access
Arailym TleubayevaArailym Tleubayeva
Arailym Tleubayeva
ASAigul K. Shaikhanova
Aigul K. Shaikhanova
BOBaurzhan Ospan
Baurzhan Ospan
ASAyan Sultan
Ayan Sultan
MAMariyam Abu
Mariyam Abu
NDNurbakyt Darmenkyzy
Nurbakyt Darmenkyzy
In this research, the focus is on recognizing 28 emotions in a text using the Roberta model, which is a state-of-the-art pre-trained language model that has achieved outstanding results in various natural language processing tasks. The study explores the effectiveness of the Roberta model for emotion recognition and compares it with other approaches, such as CNNs and RNNs. In addition, the research investigates the problem of toxicity detection, which involves identifying and flagging potentially harmful or offensive language in a given text. Various techniques for toxicity detection are considered, including supervised learning and deep learning methods. The study also explores the process of extracting key phrases and words from a text using machine learning algorithms. This involves applying NLP techniques such as part-of-speech tagging, named entity recognition, and text summarization. All of these methods are implemented and tested using a cloud service provided by Umai Cloud Services, a Kazakh startup company that offers machine learning and artificial intelligence solutions. The results of the study demonstrate the effectiveness of the Roberta model for emotion recognition and show promising results for toxicity detection and text summarization.

30 December 2022 in Scientific Journal of Astana IT University
Verified
Journal Article
Open access
PDF file
Arailym TleubayevaArailym Tleubayeva
Arailym Tleubayeva
AMAssar Maidanov
Assar Maidanov
AKArina Kantayeva
Arina Kantayeva
Traditional street lighting systems receive data about daylight levels and adjust lighting. However, in such conditions, energy consumption increases since the sensors of such systems receive data on only one indicator which is daylight. Therefore, a suitable automated intelligent lighting system model is needed. Intelligent lighting systems can adjust the brightness of the light not only based on natural data, but also based on the movement of vehicles and people. This paper describes the development, implementation, and testing of a smart lighting system model to increase energy efficiency and high reliability. This system is controlled by a micro-controller programmed to control the lighting and receive data from sensors for processing with good efficiency. Distributed sensors record environmental conditions such as daylight and traffic. Photo-resistors change resistance in daylight to light up the streets at night. The HC-SR501 infrared motion sensor detects objects emitting infrared radiation (heat) in the controlled motion zone and sends a signal to the micro-controller. The intelligent lighting system uses LED's, which consume less energy and achieve high efficiency. Calculations show that the efficiency of using these lamps is almost 70%, compared to what is used in conventional street lighting systems.