SMART EDUCATIONAL AGENT FOR LEARNING SUPPORT EMPLOYING LARGE LANGUAGE MODELS THROUGH LANGCHAIN

Abstract

The rapid development of Artificial Intelligence (AI) and Large Language Models (LLMs) has created new opportunities for providing personalized and context-aware learning support to university students. This study presents a Smart Educational Agent designed to support learning through the integration of Large Language Models with the LangChain framework. The proposed system processes educational content and stores it in a vector-based knowledge base using Qdrant, enabling relevant information to be retrieved according to students’ queries. The system provides three major learning functions: multiple-choice test generation, academic learning plan generation, and context-based question answering. Retrieval-Augmented Generation (RAG) is employed to retrieve relevant educational information and provide contextual information to the language model before generating responses. An experimental methodology was adopted, and the system was evaluated through functional laboratory tests using educational content and simulated student queries. The evaluation focused on the accuracy, contextual understanding, information retrieval, response relevance, and resource-generation capability of the proposed agent. The results demonstrated that the Smart Educational Agent was able to generate relevant multiple-choice questions, structured academic learning plans, and context-grounded responses based on the available educational content. The findings indicate that the integration of LLMs, LangChain, Qdrant, and RAG can provide an effective approach for developing intelligent learning-support systems. The proposed agent can assist students in practicing subject knowledge, organizing their learning activities, and obtaining responses based on relevant educational resources.