Cluster approach to criteria evaluation of the quality of a student’s educational outcome

• Published in: Vestnik Rossiĭskogo universiteta druzhby narodov. Serii͡a︡ Informatizat͡s︡ii͡a︡ obrazovanii͡a Vol. 19; no. 3; pp. 196 - 207
• Main Authors: Pak, Nikolai I., Klunnikova, Margarita M.
• Format: Journal Article
• Language: English
• Published: Peoples’ Friendship University of Russia (RUDN University) 15.12.2022
• Subjects: clustering of educational results; criteria assessment; quality of educational results; recognition
• ISSN: 2312-8631; 2312-864X
• DOI: 10.22363/2312-8631-2022-19-3-196-207

Problem and goal . The issues of criteria-based evaluation of the student's educational results remain relevant for the modern theory and practice of education. As a rule, measures to monitor educational results and resources in educational institutions are carried out by expert, manual, non-automated methods. In accordance with the directions of digital transformation of education, it is necessary to create a technological assessment system that meets the requirements of modern society, subject to automation and intellectualization. The purpose of the work is to substantiate a new model of criteria-based assessment of the quality of the educational result, based on the mathematical methods of the theory of clustering and pattern recognition and allowing to automate the procedures for assessing the quality of educational objects, resources, educational and personal achievements of students. Methodology. The quality of an educational result or resource is determined by criteria indicators, which can be represented as features of the evaluated object using the information vector. By clustering the set of acceptable objects into three classes - with low, medium and high quality - it is possible to evaluate an object by its belonging to one of these classes. Clustering is carried out on the basis of a mining algorithm, the metric of city blocks is taken as a measure of the similarity of objects. Results. A program has been developed that consists of a source data module, a clustering module, and a recognition and training module. The model results of the program correlate with traditional rating assessments, in which the quality of the object is determined by a point scale. The obtained test results confirm the validity of the recognition algorithm and the correctness of the software product. Conclusion. Thus, the proposed model based on clustering and the recognition method showed the possibility of automated assessment of the quality of educational results of trainees and educational resources.