A Model for Generating Tests and Exam Tickets Based on Semantic Analysis of Educational Materials

Abstract

This article analyzes the conceptual architecture and methodological foundations of an intelligent system designed to automatically generate test and exam ticket questions based on semantic analysis of educational materials. Driven by the need to automate assessment processes on digital learning platforms, the paper examines technologies for identifying the semantic structure of texts, modeling conceptual relations, and generating questions using ontological graphs and modern vector-based language models (FastText, BERT, XLM-R). The study discusses text preprocessing, semantic representation, and generating test questions based on facts, definitions, and cause–effect relations, as well as creating exam ticket questions through clustering and topic modeling. The evaluation of generated questions is presented through both automated metrics and expert criteria. The findings demonstrate that automatic question generation systems can significantly optimize assessment processes in higher education and improve the content quality of the learning process.