Özet:

Research in the field of semantic text analysis begins with the study of the structure of natural language. The Kazakh language is unique in that it belongs to agglutinative languages and requires careful study. The object of this study is the text in the Kazakh language. Existing approaches to the study of the semantic analysis of text in the Kazakh language do not consider text analysis using the methods of thematic modeling and learning of neural networks. The purpose of this study is to determine the quality of a topic model based on the LDA (Latent Dirichlet Allocation) method with Gibbs sampling, through neural network learning. The LDA model can determine the semantic probability of the keywords of a single document and give them a rating score. To build a neural network, one of the widely used LSTM architectures was used, which has proven itself well in working with NLP (Natural Language Processing). As a result of learning, it is possible to see to what extent the text was trained and how the semantic analysis of the text in the Kazakh language went. The system, developed on the basis of the LDA model and neural network learning, combines the detected keywords into separate topics. In general, the experimental results showed that the use of deep neural networks gives the expected results of the quality of the LDA model in the processing of the Kazakh language. The developed model of the neural network contributes to the assessment of the accuracy of the semantics of the used text in the Kazakh language. The results obtained can be applied in systems for processing text data, for example, when checking the compliance of the topic and content of the proposed texts (abstracts, term papers, theses, and other works). Author Biographies Akerke Аkanova, S. Seifullin Kazakh Agro Technical University PhD, Senior Lecturer Department of Computer Engineering and Software

Anahtar Kelimeler: