Аннотация
This study focuses on the classification of Kazakh text data using deep learning models. Text classification constitutes a foundational component within the domain of natural language processing (NLP), frequently serving as a preliminary step in numerous other text-processing applications. The research considered the primary task of text data classification, i.e. sentiment analysis. The study proposed the use of deep learning models for the automated classification of text data into positive and negative opinions. Since the Kazakh is a low-resource language, it is imperative to create specialised datasets, perform preprocessing of text, and carry out annotation. The study utilised the KazSAnDRA dataset, in conjunction with a dataset prepared by the authors. A range of deep learning models, including CNN, LSTM, GRU, and DistilBERT, were utilised in the experimental evaluations. Among the models examined, the DistilBERT and LSTM models demonstrated the highest levels of accuracy and were identified as a promising approach for the automatic classification of Kazakh texts. The findings of this study have the potential to provide a theoretical and practical foundation for future research endeavours focused on the analysis of texts in the Kazakh language.