Аналіз можливостей llms щодо допомоги у вивченні мови sql

Abstract

This study analyzes the potential of Large Language Models (LLMs) in assisting SQL learning. A web application was developed to test students’ SQL knowledge, integrating an automated assessment method that combines heuristic-based evaluation and GPT-based analysis. The results indicate that the GPT-based approach provides more accurate and fair assessments compared to heuristic rules, though challenges remain with complex SQL queries. Future research aims to refine the model by incorporating a cascading LLM framework to enhance evaluation accuracy and minimize inconsistencies.