Learning to Generate Structured Queries from Natural Language with Indirect Supervision

TL;DR

This paper introduces a novel approach for generating SQL queries from natural language questions using indirect supervision from answer data, leveraging reinforcement learning to improve performance across multiple domains.

Contribution

It proposes a new answer-driven learning paradigm and an end-to-end neural model that bypasses the need for explicit SQL annotations, utilizing abundant question-answer pairs.

Findings

Model outperforms baseline methods on multiple domain datasets.

Answer-driven training reduces reliance on SQL annotations.

Reinforcement learning enhances SQL generation accuracy.

Abstract

Generating structured query language (SQL) from natural language is an emerging research topic. This paper presents a new learning paradigm from indirect supervision of the answers to natural language questions, instead of SQL queries. This paradigm facilitates the acquisition of training data due to the abundant resources of question-answer pairs for various domains in the Internet, and expels the difficult SQL annotation job. An end-to-end neural model integrating with reinforcement learning is proposed to learn SQL generation policy within the answer-driven learning paradigm. The model is evaluated on datasets of different domains, including movie and academic publication. Experimental results show that our model outperforms the baseline models.