Меню
Vol. 1 No. 30 (2026) / 2026-03-30 / Number of views: 148
Authors
Keywords
Link to DOI:
How to quote
Digital platforms and environments have become an integral part of educational technologies and the learning process. Learning materials and assignments are designed and provided to students in digital format, which has made significant changes to the process of learning and acquiring knowledge. These changes concern not only digital teaching methods but also knowledge testing.
The paper presents a comprehensive AI-based SQL training support system that combines three key components: a classification of student errors by type based on the RuBERT model, an XGBoost-based recommendation module for selecting revision topics based on error profile and work pace, and adaptive testing using a two-parameter item response theory model (IRT), allowing dynamic adjustment of the complexity of questions to the current assessment of the student's ability. The system is implemented as a web application and tested on data from 120 students majoring in Information Systems, collected during their natural use of the learning application (September 2024 – May 2025): SQL logs, answers to theoretical questions, and timestamps of passing tests.