Abstract
This article presents the development of an intelligent information system for detecting near- duplicate texts in the Kazakh language based on a previously developed hybrid neural model combining statistical and semantic methods of text analysis. The relevance of the research is due to the growth of digital scientific and educational materials and the need to ensure academic integrity and automated monitoring of text borrowing in the context of the linguistic structure of the Kazakh language, characterized by high morphological variability. As part of this study, a conceptual model of the information system was developed, and a structural model of the system based on microservice architecture was formed, The database of the system monitoring of academic texts in the Kazakh language has been designed and a mechanism for integrating the hybrid model into the software environment has been implemented. Functional, integration and load testing of developed modules has been carried out. Additionally, an experimental system test was performed on examples of real-world ways to disguise text borrowings, including inserting a text fragment into an image, rephrasing, and changing the document encoding. The developed system provides automated processing of documents, fragmentation of texts, calculation of similarity ratio, detection of near-duplicates, visualization of matching fragments and formation of reports on originality degree. The results of the study confirm the correctness of calculations, system stability, and the practical applicability of the system in educational and scientific organizations.