RASL: Retrieval Augmented Schema Linking for Massive Database Text-to-SQL

TL;DR

This paper introduces RASL, a retrieval-based approach that decomposes database schemas into semantic units for scalable, accurate text-to-SQL conversion in large enterprise databases without fine-tuning.

Contribution

The paper presents a novel retrieval architecture that leverages schema decomposition and semantic indexing to improve scalability and accuracy in enterprise-level text-to-SQL systems.

Findings

Outperforms baseline methods on large, complex databases.

Maintains high recall and accuracy with extensive metadata.

Enables deployment without domain-specific fine-tuning.

Abstract

Despite advances in large language model (LLM)-based natural language interfaces for databases, scaling to enterprise-level data catalogs remains an under-explored challenge. Prior works addressing this challenge rely on domain-specific fine-tuning - complicating deployment - and fail to leverage important semantic context contained within database metadata. To address these limitations, we introduce a component-based retrieval architecture that decomposes database schemas and metadata into discrete semantic units, each separately indexed for targeted retrieval. Our approach prioritizes effective table identification while leveraging column-level information, ensuring the total number of retrieved tables remains within a manageable context budget. Experiments demonstrate that our method maintains high recall and accuracy, with our system outperforming baselines over massive databases with varying structure and available metadata. Our solution enables practical text-to-SQL systems deployable across diverse enterprise settings without specialized fine-tuning, addressing a critical scalability gap in natural language database interfaces.