Information and communication and chemical technologies

Vol. 1 No. 30 (2026) / 2026-03-30 / Number of views: 165

DESIGN OF A CLIENT-SERVER APPLICATION WITH A PERSONALIZED RECOMMENDATION SYSTEM FOR ENGLISH LANGUAGE LEARNING

Authors

Astana International University
https://orcid.org/0009-0004-5803-8674
Astana International University
https://orcid.org/0000-0001-7494-9794
Astana International University
https://orcid.org/0000-0002-6381-9350
Astana International University
https://orcid.org/0009-0000-7000-0653

Keywords

personalized recommendations, client-server application, adaptive learning, English language, system design, educational technologies, database, machine learning, XGBoost, ROC-AUC

Link to DOI:

https://doi.org/10.58805/kazutb.v.1.30-1336

How to quote

Orynbay, D.D., et al. “DESIGN OF A CLIENT-SERVER APPLICATION WITH A PERSONALIZED RECOMMENDATION SYSTEM FOR ENGLISH LANGUAGE LEARNING”. Vestnik KazUTB, vol. 1, no. 30, Mar. 2026, doi:10.58805/kazutb.v.1.30-1336.

Abstract

The article presents the design of a client-server application with personalized recommendations for learning the English language. The relevance is determined by the need to overcome the limitations of traditional methods that do not take into account students’ individual characteristics. The study investigates architectural solutions for creating individualized learning trajectories based on progress analysis and adaptation of difficulty according to the CEFR scale.

Methods of system analysis of educational platforms, object-oriented design, database normalization, and content filtering were applied. A three-tier architecture was developed: presentation layer (HTML5, CSS3, JavaScript), business logic layer (PHP), and data layer (MySQL).

A three-stage personalization methodology has been designed: an initial adaptive assessment consisting of 20 questions, automatic generation of a learning trajectory with the selection of five thematic lessons, and dynamic adaptation of learning materials. A diagram with three types of actors has been developed: an unauthenticated user, a student with access to more than 10 functions, and a teacher with progress monitoring capabilities.

The results include a scalable architecture, a methodology with automatic content selection based on scores (0-10 - A1, 11-13 - A2, 14-17 - B1, 18-20 - B2), ML comparison of three classification algorithms (Logistic Regression, XGBoost, SVM) with Precision/Recall/F1/ROC-AUC metrics, computational complexity assessment O(n·m), and integration with four external API services.