Improving Text-to-SQL with Schema Dependency Learning
This work addresses the inference speed bottleneck in text-to-SQL systems for real-world applications, offering a more flexible solution.
The paper tackles the problem of slow inference in text-to-SQL models by proposing a schema dependency learning method that reduces reliance on execution-guided decoding, achieving state-of-the-art performance with only a small performance drop while significantly speeding up inference.
Text-to-SQL aims to map natural language questions to SQL queries. The sketch-based method combined with execution-guided (EG) decoding strategy has shown a strong performance on the WikiSQL benchmark. However, execution-guided decoding relies on database execution, which significantly slows down the inference process and is hence unsatisfactory for many real-world applications. In this paper, we present the Schema Dependency guided multi-task Text-to-SQL model (SDSQL) to guide the network to effectively capture the interactions between questions and schemas. The proposed model outperforms all existing methods in both the settings with or without EG. We show the schema dependency learning partially cover the benefit from EG and alleviates the need for it. SDSQL without EG significantly reduces time consumption during inference, sacrificing only a small amount of performance and provides more flexibility for downstream applications.