INTELLIGENT TECHNOLOGY FOR ANALYSIS OF TEMPORAL ORDERING OF ELEMENTS IN THE STRUCTURAL-LOGICAL SCHEME OF AN EDUCATIONAL PROGRAM

Abstract

The article presents an intelligent technology that enables automated analysis of educational components within higher education curriculum and determines the logical sequence of their teaching based on the structuring of key concepts and life cycle processes. The technology relies on modern natural language processing (NLP) methods, particularly involving Large Language Models (LLM), which allows for identification and classification of course order in the educational process according to their content. The approach consists of 4 stages. Initially, data preparation is performed, which includes the selection of educational components relevant to defined specific competencies and learning outcomes. Next, keywords grouped by key concepts are identified in the description of topics from the syllabi of these components. In the third stage, words that reflect actions and processes related to key concepts are determined. At the final stage, a pre-trained classification AI model is used, capable of determining the stage of a multi-stage teaching process to which the discipline (component) should be assigned in the educational process. Experimental application of the developed technology for an actual bachelor's degree program in Information Systems and Technologies demonstrated high method effectiveness: 367 keywords were identified, grouped into 30 key concepts covering various aspects of the courses. A training dataset was developed for the concept of "software product," on which classification accuracy for a three-stage process (triads) of 85-87% was achieved using the gradient boosting algorithm. The created pre-trained model was successfully used to develop recommendations for the analyzed educational program. The practical value lies in the possibility of implementing this technology for analyzing structural and logical frameworks, accelerating analysis and improving the quality of development and periodic updates of educational programs. Future prospects include expanding methods for processing large text corpora, improving classification algorithms, and testing the approach in related fields of knowledge to form a universal tool for systematizing the learning process.