EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

SCDS: Data Synthesis with Structural Priors via Large Language Models for Text-to-SQL

Haoyu Zhang, Yingqi Zou, Xiangyu Wang, Shaohan Wang, Zijian Zhang

Information · 01.11.2026 · 17 (11) · 1089

DOI: 10.3390/info17111089 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Information, online erschienen am 01.11.2026. Verfasst von Haoyu Zhang und 4 weiteren Autor:innen. Die Arbeit nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: China-Developed Multimodal Large Model and Knowledge Platform for the Maritime Industry.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Text-to-SQL data synthesis must expand semantic coverage without losing the relational structure of observed queries. Existing SQL offers evidence of plausible join and nesting patterns, but reusing queries alone limits the range of training examples; unconstrained generation, in contrast, risks implausible structures and question–SQL mismatches. We introduce Structure-Constrained Data Synthesis (SCDS), which treats observed query structure as a prior and uses large language models (LLMs) to vary clause-level semantic content. SCDS groups source queries by coarse structural signatures and constructs semi-frozen SQL templates that retain the source FROM/JOIN context while abstracting selected clauses into semantic slots. It grounds intent generation in schema semantics and available values and represents the resulting objectives in a semi-structured intermediate representation. It then instantiates SQL, evaluates its executability diagnostically, and generates the corresponding questions and reasoning descriptions. On BIRD-dev, fine-tuning with the full set of 23,374 synthesized examples achieves a mean execution accuracy of 57.69% across three training seeds. On MaritimeSQLBench, fine-tuning with synthesized examples improves execution accuracy from 47.69% to 56.75% and reduces the execution error rate from 13.33% to 10.77%. The maritime result is measured with the gold schema elements required by each reference SQL. We publicly release MaritimeSQLBench to support further research on maritime Text-to-SQL. These results support structural priors as a useful constraint on Text-to-SQL synthesis. At a matched training size of 9428 examples, SCDS achieves performance comparable to BIRD-train; together with the maritime results, this highlights its potential for domain-specific Text-to-SQL data synthesis.

Förderung

  • China-Developed Multimodal Large Model and Knowledge Platform for the Maritime Industry

Zitieren

APA

Zhang, H., Zou, Y., Wang, X., Wang, S., & Zhang, Z. (2026). SCDS: Data Synthesis with Structural Priors via Large Language Models for Text-to-SQL. Information, 17(11), 1089. https://doi.org/10.3390/info17111089

Vancouver

Zhang H, Zou Y, Wang X, Wang S, Zhang Z. SCDS: Data Synthesis with Structural Priors via Large Language Models for Text-to-SQL. Information. 2026;17(11):1089. doi:10.3390/info17111089

BibTeX
@article{Zhang2026,
  author = {Zhang, Haoyu and Zou, Yingqi and Wang, Xiangyu and Wang, Shaohan and Zhang, Zijian},
  title = {SCDS: Data Synthesis with Structural Priors via Large Language Models for Text-to-SQL},
  journal = {Information},
  year = {2026},
  volume = {17},
  number = {11},
  pages = {1089},
  doi = {10.3390/info17111089}
}
Kostenlos starten →