Uso de LLM e RAG para construção de um chatbot voltado a potenciais turistas e moradores de João Pessoa

Abstract

This work presents the development of a prototype conversational agent (chatbot) based on a Large Language Model (LLM), incorporating advanced Retrieval-Augmented Generation (RAG) techniques to enhance its performance. The agent was designed to operate within the context of Jo˜ ao Pessoa, a city in the state of Para´ ıba, Brazil, which was chosen due to its growing national prominence in tourism and quality of life. This makes the city an excellent case study for implementing an LLM-based conversational agent. The chatbot demonstrated strong conversational abilities according to the proposed evaluation metrics, indicating its potential as a means of promoting the city.