RAG-BASED CONVERSATIONAL AGENT FOR COLLEGE WEBSITE NAVIGATION

Abstract

Educational institutions often face challenges in providing timely and accurate responses to numerous student queries related to admissions, courses, fee structures, and campus policies. Traditional communication methods such as emails and help desks are limited in scalability and efficiency, resulting in delays and a poor user experience. Addressing these limitations requires an automated, intelligent solution capable of understanding natural language queries and delivering reliable, context-based information.

This project introduces an AI-powered college chatbot that combines Large Language Models (LLM) with Retrieval-Augmented Generation (RAG) to provide accurate and conversational responses. The system retrieves information from verified sources, such as the college website and official documents, reducing hallucinations and ensuring data reliability. It employs a

structured data pipeline involving web scraping, PDF parsing, embedding generation, and semantic search using a vector database. Key features include multi-language support, voice interaction, and secure authentication for personalized queries. The chatbot is deployed on the official college website, delivering a scalable and efficient solution that enhances accessibility, accuracy, and overall user experience in academic information services.