Дослідження питання використання llms для генерації тестів з метою моніторингу розуміння прочитаного матеріалу

Abstract

This paper is devoted to the study of using LLMs to generate tests based on book texts to monitor reading understanding. Five models were chosen to select the best one, tests were generated, and given to students to take. After passing each test, students left their scores for this test. Evaluation results led to the selection of the optimal model, considering scores, price, and generation speed. Based on the selected model, further work will be done to develop a reading application with tests generation, tests execution and control of reading understanding.