![[2508.07087] SQL-Exchange: Transforming SQL Queries Across Domains](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2508.07087] SQL-Exchange: Transforming SQL Queries Across Domains
Summary
SQL-Exchange introduces a framework for transforming SQL queries across different database schemas while maintaining structural integrity, enhancing text-to-SQL performance.
Why It Matters
As databases become increasingly diverse, the ability to adapt SQL queries across different schemas is crucial for improving data accessibility and usability. SQL-Exchange addresses this challenge, potentially streamlining workflows in data science and AI applications.
Key Takeaways
- SQL-Exchange enables effective mapping of SQL queries across diverse database schemas.
- The framework preserves source query structure while adapting domain-specific elements.
- In-context prompting and fine-tuning with mapped queries improve text-to-SQL performance.
- Comprehensive evaluation shows effectiveness across various schemas and query types.
- The approach enhances the usability of text-to-SQL systems in real-world applications.
Computer Science > Databases arXiv:2508.07087 (cs) [Submitted on 9 Aug 2025 (v1), last revised 22 Feb 2026 (this version, v2)] Title:SQL-Exchange: Transforming SQL Queries Across Domains Authors:Mohammadreza Daviran, Brian Lin, Davood Rafiei View a PDF of the paper titled SQL-Exchange: Transforming SQL Queries Across Domains, by Mohammadreza Daviran and 2 other authors View PDF HTML (experimental) Abstract:We introduce SQL-Exchange, a framework for mapping SQL queries across different database schemas by preserving the source query structure while adapting domain-specific elements to align with the target schema. We investigate the conditions under which such mappings are feasible and beneficial, and examine their impact on enhancing the in-context learning performance of text-to-SQL systems as a downstream task. Our comprehensive evaluation across multiple model families and benchmark datasets -- assessing structural alignment with source queries, execution validity on target databases, and semantic correctness -- demonstrates that SQL-Exchange is effective across a wide range of schemas and query types. Our results further show that both in-context prompting with mapped queries and fine-tuning on mapped data consistently yield higher text-to-SQL performance than using examples drawn directly from the source schema. Comments: Subjects: Databases (cs.DB); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) Cite as: arXiv:2508.07087 [cs.DB] (or arXiv:2508.07...
