Informações Principais
     Resumo
     Abstract
     Introdução
     Conclusão
     Download
  
  
  
 
Abstract
 
 
Acadêmico(a): Hélinton Pereira Steffens
Título: ADA: ASSISTENTE PESSOAL AUTOMATIZADO BASEADO EM LINGUAGEM NATURAL
 
Abstract:
This monograph presents the development of an automated personal assistant to help users manage their activities, being able to converse in Portuguese and English, answering and/or understanding questions and user statements. Upon receiving a sentence, the assistant verifies if the interaction was carried out by text or voice. If it is by voice, the system translates the speech into text, using an Automatic Speech Recognition (ASR) library, and identifies the language of the interaction. Then, the sentence is analyzed by the Natural Language Processing (PLN) to extract subject, verb and complements, as well as identify if it is a statement, a question or a negation. The statements are stored in a graph format by the semantic web, thus allowing the user to perform queries and searches (questions), using rules of deduction and reasoning. The answers to questions are inferred from user knowledge models and can be returned by voice or text. In case the output is voice, the assistant transforms generated responses from text to speech using the Text To Speech (TTS) library. The proposed work reached the initials objectives, with the restriction that the phrases required need to have subject and verb and optionally complements.