Application of Kohonen Self-organizing Maps for Balancing Test Questions in E-Learning

The paper concerns the issue of clustering and balance of test questions on knowledge, skills and abilities of university students. Opportunities of developing nowadays electronic learning has a great number of advantages in providing learning material and control of obtained knowledge, skills and abilities of students over traditional methods and means: flexibility, accessibility, cost-effectiveness, interactivity, multimedia, individualization and differentiation of learning, high degree of students’ involvement. Formation of knowledge, skills and abilities in students is the most important process in learning. The level of knowledge, skills and abilities determines the success of university students in their future profession. That is why at the stage of designing electronic tests, it is important for the teacher to develop questions based on what the student should know and be able to do, what skills to possess. One of the main functions of the learning management system Moodle, as an e-learning tool, is the diagnostic function. LMS Moodle allows to optimize and automate the process of controlling the mastery of learning material. Thanks to this environment the teacher has the opportunity to easily develop test questions of open and closed types. However, in order for these test questions and e-testing in general to become effective, valid and reliable, it is very important for the teacher to keep the balance of test questions in terms of their categories, focus on testing knowledge, skills and abilities. Each category should contain a roughly equal number of questions so that e-testing can provide a complete view of student learning outcomes. The study involved the use of Kohonen self-organizing maps for clustering of test questions, which allowed to assess the balance of available questions in terms of knowledge, skills, and abilities and to make adjustments in the bank of test questions to achieve a high degree of efficiency and objectivity of e-testing.