Faster and Better Grammar-based Text-to-SQL Parsing via Clause-level Parallel Decoding and Alignment Loss

TL;DR

This paper introduces clause-level parallel decoding and alignment loss to improve the efficiency and accuracy of grammar-based text-to-SQL parsers, achieving faster decoding and better performance.

Contribution

It proposes novel decoding and alignment techniques that enhance existing parsers, RATSQL and LGESQL, in both speed and accuracy.

Findings

Significant speedup in decoding process

Improved accuracy in cross-domain text-to-SQL parsing

Consistent performance gains across two parsers

Abstract

Grammar-based parsers have achieved high performance in the cross-domain text-to-SQL parsing task, but suffer from low decoding efficiency due to the much larger number of actions for grammar selection than that of tokens in SQL queries. Meanwhile, how to better align SQL clauses and question segments has been a key challenge for parsing performance. Therefore, this paper proposes clause-level parallel decoding and alignment loss to enhance two high-performance grammar-based parsers, i.e., RATSQL and LGESQL. Experimental results of two parsers show that our method obtains consistent improvements both in accuracy and decoding speed.