# Clause-Wise and Recursive Decoding for Complex and Cross-Domain Text-to-SQL Generation

## TL;DR

This paper introduces a clause-wise and recursive neural architecture with self-attention for complex, cross-domain text-to-SQL generation, significantly improving accuracy on the Spider dataset, especially for nested queries.

## Contribution

The paper presents a novel clause-wise decoding neural model with recursive capabilities and a self-attention schema encoder for the challenging Spider dataset.

## Key findings

- Achieves 4.6% and 9.8% accuracy improvements on test and dev sets.
- More effective at predicting complex and nested SQL queries.
- Outperforms previous models on the Spider dataset.

## Abstract

Most deep learning approaches for text-to-SQL generation are limited to the WikiSQL dataset, which only supports very simple queries over a single table. We focus on the Spider dataset, a complex and cross-domain text-to-SQL task, which includes complex queries over multiple tables. In this paper, we propose a SQL clause-wise decoding neural architecture with a self-attention based database schema encoder to address the Spider task. Each of the clause-specific decoders consists of a set of sub-modules, which is defined by the syntax of each clause. Additionally, our model works recursively to support nested queries. When evaluated on the Spider dataset, our approach achieves 4.6\% and 9.8\% accuracy gain in the test and dev sets, respectively. In addition, we show that our model is significantly more effective at predicting complex and nested queries than previous work.