Representing Schema Structure with Graph Neural Networks for Text-to-SQL Parsing

TL;DR

This paper introduces a graph neural network-based method to encode database schema structures for text-to-SQL parsing, significantly improving accuracy on complex schemas in the Spider dataset.

Contribution

The paper presents a novel approach that encodes database schema structure with graph neural networks, enhancing parser performance on complex, unseen schemas.

Findings

Schema structure encoding improves accuracy from 33.8% to 39.4%.

Achieves state-of-the-art performance at 39.4%.

Demonstrates the importance of schema structure in text-to-SQL parsing.

Abstract

Research on parsing language to SQL has largely ignored the structure of the database (DB) schema, either because the DB was very simple, or because it was observed at both training and test time. In Spider, a recently-released text-to-SQL dataset, new and complex DBs are given at test time, and so the structure of the DB schema can inform the predicted SQL query. In this paper, we present an encoder-decoder semantic parser, where the structure of the DB schema is encoded with a graph neural network, and this representation is later used at both encoding and decoding time. Evaluation shows that encoding the schema structure improves our parser accuracy from 33.8% to 39.4%, dramatically above the current state of the art, which is at 19.7%.