Large Language Model-Powered Retrieval-Augmented Generation System for Academic Knowledge Management

Abstract

LLMs have transformed the field of Artificial Intelligence and NLP by enabling conversational systems capable of understanding and generating human-like responses. In recent years, students and university staff have increasingly relied on AI systems to obtain information quickly and efficiently rather than spending long periods navigating complex university websites and administrative platforms. The new generation of users expects immediate access to information and often prefers asking intelligent assistants directly instead of manually searching through multiple webpages, documents, and announcements. However, general-purpose AI models frequently produce inaccurate or hallucinated answers because they lack direct access to institution-specific data and are not continuously synchronized with university resources and updates. This project addresses this problem through the development of an LLM-Powered RAG System for Academic Knowledge Management. The primary objective of the proposed system is to provide students, teachers, researchers, and university staff with a reliable intelligent assistant capable of retrieving accurate academic and administrative information directly from official university websites and digital resources. Unlike AI chatbots that rely solely on pretrained knowledge, the proposed solution dynamically retrieves relevant information from university data sources before generating responses, thereby improving factual accuracy, contextual relevance, and institutional reliability.