Interacting with Student Information on Google Sheets Using the RAG Technique

Abstract

This study presents an AI-powered chatbot that enables natural language access to student records stored in Google Sheets. The system integrates Retrieval-Augmented Generation (RAG) using Gemini 1.5 Flash and ChromaDB for semantic embedding, vector-based retrieval, and contextual response generation. A RESTful API, developed using Python Flask, supports integration with external platforms, allowing users to query academic, health, contact, and demographic data. Developed under the Design and Development Research (DDR), Type I model, the prototype was tested with 15 representative queries against a mock dataset of 35 students. Results showed average token usage of $\mathbf{3 1 6. 5}$ for prompts and $\mathbf{3 7. 7}$ for responses, with 14 of 15 answers receiving the highest relevance score. The findings demonstrate that a lightweight, RAG-based chatbot can offer efficient and accurate educational data access, supporting non-technical users in retrieving information without direct spreadsheet interaction.

Ähnliche Arbeiten