CHATBOT RAG PERSONALIZADO DE DADOS EM MODELOS DE LINGUAGEM NATURAL GENERATIVOS

Abstract

This work presents the development of a chatbot based on Retrieval-Augmented Gene ration (RAG), adapted to process data from the official announcements of IF Goiano Campus Urutaí, specifically those related to research, graduation regulations, and student assistance. The system aims to optimize query time for students and professors and reduce the adminis trative workload of secretariats and coordinations. The development methodology was divided into five phases: data collection and preparation, creation of the knowledge base with embed dings, RAG system development, construction of the user interface (chatbot), and finally, sys tem testing and evaluation. After several tests and database improvements, positive results were achieved, demonstrating the solution’s viability, although potential for future improvements and implementations was also identified.