In recent years we have seen the growth of the volume of unstructured data generated in the traditional Web. Therefore the Semantic Web was born as a paradigm that proposes to structure the content of the Web flexibly through domain ontologies and the RDF model, making computers capable of automatically processing this data, enabling the generation of more information and knowledge. However, to make this information accessible to users in other domains, there needs to be a more convenient way of looking at these knowledge bases. The Natural Language Processing (NLP) area has provided tools to allow natural (spoken or writing) is a convenient way to perform queries in knowledge bases. However, for the use of natural language to be useful, a method is required that converts a natural language question or request into a structured query. With this objective, the present work proposes an approach that converts a question/request in Portuguese into a structured query in the SPARQL language, through the use of dependency trees and structured ontologies in graphs, and that also enables the enrichment of question/request results by generating related questions.