Saved in:
Bibliographic Details
Main Authors: Srikanth, Neha, Bursztyn, Victor, Mathur, Puneet, Nenkova, Ani
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.27532
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912679859322880
author Srikanth, Neha
Bursztyn, Victor
Mathur, Puneet
Nenkova, Ani
author_facet Srikanth, Neha
Bursztyn, Victor
Mathur, Puneet
Nenkova, Ani
contents We introduce SQLSpace, a human-interpretable, generalizable, compact representation for text-to-SQL examples derived with minimal human intervention. We demonstrate the utility of these representations in evaluation with three use cases: (i) closely comparing and contrasting the composition of popular text-to-SQL benchmarks to identify unique dimensions of examples they evaluate, (ii) understanding model performance at a granular level beyond overall accuracy scores, and (iii) improving model performance through targeted query rewriting based on learned correctness estimation. We show that SQLSpace enables analysis that would be difficult with raw examples alone: it reveals compositional differences between benchmarks, exposes performance patterns obscured by accuracy alone, and supports modeling of query success.
format Preprint
id arxiv_https___arxiv_org_abs_2510_27532
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SQLSpace: A Representation Space for Text-to-SQL to Discover and Mitigate Robustness Gaps
Srikanth, Neha
Bursztyn, Victor
Mathur, Puneet
Nenkova, Ani
Computation and Language
We introduce SQLSpace, a human-interpretable, generalizable, compact representation for text-to-SQL examples derived with minimal human intervention. We demonstrate the utility of these representations in evaluation with three use cases: (i) closely comparing and contrasting the composition of popular text-to-SQL benchmarks to identify unique dimensions of examples they evaluate, (ii) understanding model performance at a granular level beyond overall accuracy scores, and (iii) improving model performance through targeted query rewriting based on learned correctness estimation. We show that SQLSpace enables analysis that would be difficult with raw examples alone: it reveals compositional differences between benchmarks, exposes performance patterns obscured by accuracy alone, and supports modeling of query success.
title SQLSpace: A Representation Space for Text-to-SQL to Discover and Mitigate Robustness Gaps
topic Computation and Language
url https://arxiv.org/abs/2510.27532