Learning Contextual Representations for Semantic Parsing with Generation-Augmented Pre-Training
This work provides a strong specific gain for the text-to-SQL semantic parsing community by improving the ability of models to handle complex SQL queries and schema interactions.
This paper addresses the limitations of general-purpose language models in text-to-SQL semantic parsing, specifically their inability to detect column mentions, infer them from cell values, and compose complex SQL queries. The authors introduce Generation-Augmented Pre-training (GAP), a framework that uses generative models to create 2M utterance-schema pairs and 30K utterance-schema-SQL triples for pre-training. Semantic parsers using GAP MODEL achieve new state-of-the-art results on the SPIDER and CRITERIA-TO-SQL benchmarks.
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.