Reflective Reasoning for SQL Generation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Mohr, Isabelle, Gandarela, Joao, Dujany, John, Freitas, Andre
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917194599759872
author Mohr, Isabelle
Gandarela, Joao
Dujany, John
Freitas, Andre
author_facet Mohr, Isabelle
Gandarela, Joao
Dujany, John
Freitas, Andre
contents Robust text-to-SQL over complex, real-world databases remains brittle even with modern LLMs: iterative refinement often introduces syntactic and semantic drift, corrections tend to be non-transferable across queries, and naive use of large context windows scales poorly. We propose a controlled text-to-SQL framework built around reflective refinement. Instead of repeatedly rewriting the current SQL instance, the system decomposes generation into typed stages and applies feedback as persistent updates to the stage-level generation mechanism. A Reflection-Refinement Loop localizes violations to the responsible stage maximize preservation of previously validated constraints and support monotonic improvement over a query set. The method operates without gold SQL by combining interpreter-based checks with LLM-based semantic coverage verification as epistemic judges. Experiments on Spider and BIRD demonstrate consistent gains over strong prompting baselines, robust convergence within a small refinement budget, and improved execution accuracy across both frontier and open-weight model families.
format Preprint
id arxiv_https___arxiv_org_abs_2601_06678
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Reflective Reasoning for SQL Generation
Mohr, Isabelle
Gandarela, Joao
Dujany, John
Freitas, Andre
Databases
Robust text-to-SQL over complex, real-world databases remains brittle even with modern LLMs: iterative refinement often introduces syntactic and semantic drift, corrections tend to be non-transferable across queries, and naive use of large context windows scales poorly. We propose a controlled text-to-SQL framework built around reflective refinement. Instead of repeatedly rewriting the current SQL instance, the system decomposes generation into typed stages and applies feedback as persistent updates to the stage-level generation mechanism. A Reflection-Refinement Loop localizes violations to the responsible stage maximize preservation of previously validated constraints and support monotonic improvement over a query set. The method operates without gold SQL by combining interpreter-based checks with LLM-based semantic coverage verification as epistemic judges. Experiments on Spider and BIRD demonstrate consistent gains over strong prompting baselines, robust convergence within a small refinement budget, and improved execution accuracy across both frontier and open-weight model families.
title Reflective Reasoning for SQL Generation
topic Databases
url https://arxiv.org/abs/2601.06678