SpotIt: Evaluating Text-to-SQL Evaluation with Formal Verification

TL;DR

SpotIt introduces a formal verification approach to evaluate Text-to-SQL systems, revealing that traditional test-based methods may overlook differences between generated and ground-truth queries, thus providing a more reliable assessment.

Contribution

The paper presents a novel formal verification pipeline for Text-to-SQL evaluation, extending existing verifiers to support richer SQL subsets and improving evaluation reliability.

Findings

Test-based evaluation often misses differences between queries.

Formal verification uncovers more nuanced differences in system performance.

Current evaluation methods may overestimate the accuracy of Text-to-SQL models.

Abstract

Community-driven Text-to-SQL evaluation platforms play a pivotal role in tracking the state of the art of Text-to-SQL performance. The reliability of the evaluation process is critical for driving progress in the field. Current evaluation methods are largely test-based, which involves comparing the execution results of a generated SQL query and a human-labeled ground-truth on a static test database. Such an evaluation is optimistic, as two queries can coincidentally produce the same output on the test database while actually being different. In this work, we propose a new alternative evaluation pipeline, called SpotIt, where a formal bounded equivalence verification engine actively searches for a database that differentiates the generated and ground-truth SQL queries. We develop techniques to extend existing verifiers to support a richer SQL subset relevant to Text-to-SQL. A performance evaluation of ten Text-to-SQL methods on the high-profile BIRD dataset suggests that test-based methods can often overlook differences between the generated query and the ground-truth. Further analysis of the verification results reveals a more complex picture of the current Text-to-SQL evaluation.