A Method Based on Large Language Models for Prompt Adjustment of an Individual Curriculum During E-Practice in Large-Scale Network Companies

This paper explores the objective of improving the processes of managing a student's personalized learning path (PLP) in the process of mastering a training course. It proposes a hypothesis that, in e-learning, the issue of promptly adjusting an individual curriculum during e-practice may be solved by using a specialized multiagent neural network model. To test this hypothesis, the authors have developed a multiagent model that includes agents based on large language models. The model efficiency was assessed using statistical analysis. The paper substantiates the principal provisions of the method for adjusting an individual curriculum during e-practice for intellectual support of practical student work, which, by formalizing the process of agent interaction and taking into account the impact of negative factors on the quality and time of completing the course, ensures a dynamic development of the learning path. A computational experiment using the example of a course in programming showed that applying the method in a resultcentered learning environment resulted in a 23% to 35% decrease in the time needed to achieve the required competency level.