TY - GEN
T1 - Integrating Question Answering and Text-to-SQL in Portuguese
AU - José, Marcos Menon
AU - José, Marcelo Archanjo
AU - Mauá, Denis Deratani
AU - Cozman, Fábio Gagliardi
N1 - Funding Information:
Acknowledgment. This work was partly supported by Itaú Unibanco S.A. through the Programa de Bolsas Itaú (PBI) of the Centro de Ciência de Dados da Universidade de São Paulo (C2D-USP); by the Center for Artificial Intelligence (C4AI) through support from the São Paulo Research Foundation (FAPESP grant #2019/07665-4) and from the IBM Corporation; by CNPq grants no. 312180/2018-7 and 304012/2019-0, and CAPES Finance Code 001.
Publisher Copyright:
© 2022, Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - Deep learning transformers have drastically improved systems that automatically answer questions in natural language. However, different questions demand different answering techniques; here we propose, build and validate an architecture that integrates different modules to answer two distinct kinds of queries. Our architecture takes a free-form natural language text and classifies it to send it either to a Neural Question Answering Reasoner or a Natural Language parser to SQL. We implemented a complete system for the Portuguese language, using some of the main tools available for the language and translating training and testing datasets. Experiments show that our system selects the appropriate answering method with high accuracy (over 99%), thus validating a modular question answering strategy.
AB - Deep learning transformers have drastically improved systems that automatically answer questions in natural language. However, different questions demand different answering techniques; here we propose, build and validate an architecture that integrates different modules to answer two distinct kinds of queries. Our architecture takes a free-form natural language text and classifies it to send it either to a Neural Question Answering Reasoner or a Natural Language parser to SQL. We implemented a complete system for the Portuguese language, using some of the main tools available for the language and translating training and testing datasets. Experiments show that our system selects the appropriate answering method with high accuracy (over 99%), thus validating a modular question answering strategy.
KW - Natural language interfaces to databases
KW - Natural language processing in portuguese
KW - Question answering
KW - Transformers networks
UR - http://www.scopus.com/inward/record.url?scp=85127206125&partnerID=8YFLogxK
UR - https://www.mendeley.com/catalogue/5d90488a-3bf1-39e2-9cb8-905715b926fe/
U2 - 10.1007/978-3-030-98305-5_26
DO - 10.1007/978-3-030-98305-5_26
M3 - Contribución a la conferencia
AN - SCOPUS:85127206125
SN - 9783030983048
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 278
EP - 287
BT - Computational Processing of the Portuguese Language - 15th International Conference, PROPOR 2022, Proceedings
A2 - Pinheiro, Vládia
A2 - Gamallo, Pablo
A2 - Amaro, Raquel
A2 - Scarton, Carolina
A2 - Batista, Fernando
A2 - Silva, Diego
A2 - Magro, Catarina
A2 - Pinto, Hugo
PB - Springer Science and Business Media Deutschland GmbH
T2 - 15th International Conference on the Computational Processing of Portuguese, PROPOR 2022
Y2 - 21 March 2022 through 23 March 2022
ER -