Keywords
constraint-aware reranking
schema drift
paired simulation
reproducibility
Abstract
We evaluated constraint-aware reranking for text-to-SQL execution under 37% schema drift. A deterministic paired simulation generated 56 cases and preserved a rare-condition slice. Mean execution accuracy changed from 0.499 to 0.539; the paired difference was +0.040 (95% interval +0.038 to +0.043). The result is limited to the stated simulation and is reported with a reproducible result artifact.
References
Xie, H., Zhou, X., Yang, J., Shen, S., Wang, Z., Zheng, Y., Xu, T., Shi, Y., Zong, Z., Li, Y., Chen, P., Jiang, J., He, D., Yan, X., & Jiang, J. (2026). SiriusDeliver: Automating Data Warehouse Delivery at Tencent. arXiv. https://doi.org/10.48550/arXiv.2608.09185
Furniss, P., & Green, A. (2023). SQL/PGQ data model and graph schema. https://doi.org/10.54285/ldbc.qzsk3559
Wang, Y., Chen, Y., Chen, R., Shu, H., Liu, P., & Xu, W. (2026). Enhanced Financial Text-to-SQL Generation via Fine-Grained SQL Refinement. https://doi.org/10.2139/ssrn.6502095
Beckmann, S., Wiesner, K., Tebruegge, C., & Grum, M. (2026). Combining Text-to-SQL and Large Language Models for Maintenance Decision Support. https://doi.org/10.2139/ssrn.7103027
Chafik, S., Ezzini, S., & Berrada, I. (2025). Enhancing security in text-to-SQL systems: A novel dataset and agent-based framework. Natural Language Processing, 31(6), 1399-1422. https://doi.org/10.1017/nlp.2025.10008
Guo, M., & Sun, Y. (2026). A COMPARATIVE STUDY OF LLM - POWERED DATABASE INTERFACES VERSUS TRADITIONAL SQL SYSTEMS FOR INVENTORY MANAGEMENT. NLP & Text Mining, 11-21. https://doi.org/10.5121/csit.2026.160602
Utomo, M. N. Y. (2020). Pengembangan Model Migrasi Database Relational ke NoSQL Memanfaatkan Metadata SQL. Jurnal Teknologi Elekterika, 4(2), 1. https://doi.org/10.31963/elekterika.v4i2.2212
Hairan, B., & Şahman, M. A. (2026). A comparative evaluation of large language models for detecting SQL injection vulnerabilities in web applications. PeerJ Computer Science, 12, e4015. https://doi.org/10.7717/peerj-cs.4015
Öztürk, E. (2025). Improving Text-to-Sql Conversion for Low-Resource Languages Using Large Language Models. Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, 14(1), 163-178. https://doi.org/10.17798/bitlisfen.1561298
Cinquin, O. (2024). ChIP-GPT: a managed large language model for robust data extraction from biomedical database records. Briefings in Bioinformatics, 25(2), bbad535. https://doi.org/10.1093/bib/bbad535