Learning Contextual Representations for Semantic Parsing with Generation-Augmented Pre-Training

TL;DR

This paper introduces Generation-Augmented Pre-training (GAP), a novel approach that enhances semantic parsing by jointly learning representations of language and schemas through generation models, leading to state-of-the-art results on text-to-SQL benchmarks.

Contribution

The paper proposes GAP, a pre-training framework that uses generation models to improve contextual representations for semantic parsing, addressing limitations of existing language models.

Findings

GAP achieves state-of-the-art results on SPIDER and CRITERIA-TO-SQL benchmarks.

GAP effectively detects column mentions and composes complex SQL queries.

Pre-training on large-scale generated data enhances semantic parser performance.

Abstract

Most recently, there has been significant interest in learning contextual representations for various NLP tasks, by leveraging large scale text corpora to train large neural language models with self-supervised learning objectives, such as Masked Language Model (MLM). However, based on a pilot study, we observe three issues of existing general-purpose language models when they are applied to text-to-SQL semantic parsers: fail to detect column mentions in the utterances, fail to infer column mentions from cell values, and fail to compose complex SQL queries. To mitigate these issues, we present a model pre-training framework, Generation-Augmented Pre-training (GAP), that jointly learns representations of natural language utterances and table schemas by leveraging generation models to generate pre-train data. GAP MODEL is trained on 2M utterance-schema pairs and 30K utterance-schema-SQL triples, whose utterances are produced by generative models. Based on experimental results, neural semantic parsers that leverage GAP MODEL as a representation encoder obtain new state-of-the-art results on both SPIDER and CRITERIA-TO-SQL benchmarks.