Ontology of computer-based testing in learning

The article considers the scientific and methodological foundations of computer-based testing in learning. The structure of tests in training and assessment exercises is defined and substantiated based on didactic requirements. Typical scenarios and screen forms of using training and assessment exercises are described, with examples from an electronic textbook and the publisher's electronic library system. It is shown that tests should be structured and grouped into sets according to the levels of assimilation of educational material. Recommendations are given for the preparation and use of tests at each level of assimilation, along with examples of tests with selective answers. Algorithms for computer-based testing with both fixed and variable numbers of tests in a set presented to students are proposed. The expediency of using "tickets" – predefined test sets - for final assessments is shown. A preliminary evaluation of modern neural networks' capabilities in generating computer-based tests is conducted, and recommendations for their application are provided. The problem of fraud in computer-based testing, particularly through the use of neural network "cheat sheets" is discussed. The use of electronic tutors is proposed, where computer-based testing serves as a diagnostic assessment tool in the learning process, offering individualized support to each student.