Influência da Quantidade e Abrangência em Sistemas RAG para Gestão do Espectro

Radio frequency spectrum management requires analyzing vast documentation, challenging regulatory efficiency. As a solution, this work investigates the implementation of an AI assistant based on Retrieval-Augmented Generation (RAG) within the Brazilian telecommunications regulatory context. The focus is on analyzing how the quantity and scope of data inserted into the RAG impact the quality of the responses provided. Thus, the resilience of a traditional RAG architecture is tested with the addition of correlated information and thematic noise. The results demonstrate the stability of response quality to the addition of data and noise, validating the robustness of the tested RAG architecture.