arXiv cs.CL· Ekata Mitra, Yufei Tao, Kamalamalini Nagasundaram, Ameeta Agrawal·· 11 小时前AI 评分31
GRIPS：面向 SQL 生成的预 SQL 接地信息实证研究
GRIPS: An Empirical Study of Pre-SQL Grounding Information for SQL Generation
AI 导读
GRIPS 是一种三阶段方法，在生成 SQL 前显式导出结构化预 SQL 接地信息，在 TempTabQA-C 和 BIRD 上，GRIPS-Adaptive 在 GPT-5-nano 与 Llama-70B 的 20 个设置中有 18 个优于 SQL Static，平均准确率提升 11.8 个百分点，模型调用 token 用量减少 47.8%。
正文 · 原文
Abstract:Text-to-SQL systems answer natural language questions over structured data, but can fail before SQL generation by linking question mentions to the wrong table, column, value, or relation. Even valid SQL can then return an incorrect answer. This raises a central question: What pre-SQL information helps a SQL generator, and when does additional information cease to help? We investigate this question on TempTabQA-C and BIRD. Among the audited SQL Static failures, grounding errors form the largest category on both benchmarks. We introduce GRIPS (Grounding Representation and Input Processing for SQL generation), a three-stage method that derives structured pre-SQL information to make these decisions explicit before query generation. Across GPT-5-nano and Llama-70B, GRIPS-Adaptive improves over SQL Static in 18 of 20 settings, with an average accuracy gain of 11.8 percentage points, while reducing model-call token usage by 47.8%. Our controlled analysis shows which grounding information can improve SQL generation, whereas adding more pre-SQL information does not consistently help. By separating question-input grounding from SQL generation, GRIPS makes text-to-SQL more accurate, efficient, and easier to diagnose.
| Comments: | Accepted at PACLIC 40 (2026) |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.02314 [cs.CL] |
| (or arXiv:2611.02314v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.02314 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ekata Mitra [view email]
[v1]
Mon, 2 Nov 2026 00:03:50 UTC (1,430 KB)
来源：arXiv cs.CL · arxiv.org