An Extensible Software Architecture for Intelligent Assistant


Atasoy M. E., KOÇYİĞİT A.

6th International Conference on Computer Science and Engineering, UBMK 2021, Ankara, Turkey, 15 - 17 September 2021, pp.92-97 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/ubmk52708.2021.9558940
  • City: Ankara
  • Country: Turkey
  • Page Numbers: pp.92-97
  • Keywords: Artificial intelligence, Automated speech recognition, Intelligent personal assistant, Natural language processing
  • Middle East Technical University Affiliated: Yes

Abstract

© 2021 IEEEThis article introduces an intelligent assistant architecture that allows replacing existing modules and plugging new modules without requiring any code changes thanks to its modular and extensible structure. Our proposal supports multiple natural languages and at the same time allows the integration of any external artificial intelligence or loT component easily. The software that we developed understands the commands in English and Turkish languages and transmits the answers to the user in these languages. Through our domain-specific language, different artificial intelligence components can be integrated into the system. Thanks to this domain specific language, only specific conversations are detected and the artificial intelligence module can be called. However, human speech contains many synonyms and we cannot always achieve the exact word by word. For this reason, we offer a system that understands the meanings of sentences and learns what the user wants to do when necessary.