CREATING A WORKING SOFTWARE TOOL BASED ON A MINI LLM FOR THE ADMISSION COMMITTEE

Abstract

It is noted that the admission process to higher education institutions is becoming increasingly complex from year to year, which is associated with a large number of administrative tasks, repetitive questions and increased activity of applicants. Therefore, this thesis proposes a conceptual model of an intelligent system based on Mini-LLM, designed to optimize the work of the admissions committee and automate a number of processes. This approach involves integrating Mini-LLM with a knowledge base containing admission rules, curricula, internal university procedures and relevant regulatory documents. Mini-LLM requires significantly fewer resources and funds compared to full-scale LLM models, which, while increasing the efficiency of the admission process, presents the system as a cost-effective and quickly implemented solution.