Integrating question answering and text-to-SQL in Portuguese

TL;DR

This paper presents a modular system for Portuguese that integrates question answering and text-to-SQL techniques, accurately classifying queries to improve natural language understanding and database interaction.

Contribution

It introduces a novel architecture that combines question answering and SQL parsing modules for Portuguese, validated with high accuracy.

Findings

System achieves over 99% accuracy in query classification

Effective integration of QA and text-to-SQL modules for Portuguese

Validates modular question answering strategy

Abstract

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.