An Intelligent Retrieval Augmented Generation Chatbot for Contextually-Aware Conversations to Guide High School Students

Abstract

The growth of Large Language Models (LLM) has showcased both their potential and limitations in domain- specific applications. A significant limitation is their tendency to confidently present false information, known as hallucination, which compromises the integrity of their outputs. Retrieval Augmented Generation (RAG) systems mitigate this by enhancing the Large Language Models' ability to retrieve accurate, source-specific knowledge. Many American students begin preparing for college applications in middle school, seeking guidance on planning their high school years to achieve personal goals. The diversity in students' interests, skills, and circumstances often results in an overwhelming amount of information and advice. This research introduces a novel RAG chatbot designed to provide personalized assistance to high school students using their course guide as the primary source document. By integrating a high school course guide with a RAG system, the chatbot delivers relevant suggestions, helping students navigate their educational paths with greater confidence and precision. This system aims to minimize the risk of hallucination by ensuring that responses are grounded in the accurate and relevant information contained within the course guide. Consequently, it addresses the crucial need for personalized educational planning, enabling students to make informed decisions that align with their unique aspirations and circumstances. This research highlights the transformative potential of RAG systems in educational contexts, offering a reliable tool to support students in their academic and personal growth.

Topics & Concepts