MT-Teql: Evaluating and Augmenting Consistency of Text-to-SQL Models with Metamorphic Testing

TL;DR

MT-Teql is a metamorphic testing framework that evaluates and improves the consistency of text-to-SQL models against linguistic and schema variations, significantly reducing errors and enhancing robustness.

Contribution

The paper introduces a model-agnostic metamorphic testing framework for systematically evaluating and augmenting text-to-SQL models' consistency, exposing errors, and boosting robustness.

Findings

Exposed thousands of prediction errors in SOTA models.

Enriched datasets by an order of magnitude.

Reduced over 40% inconsistency errors without accuracy loss.

Abstract

Text-to-SQL is a task to generate SQL queries from human utterances. However, due to the variation of natural language, two semantically equivalent utterances may appear differently in the lexical level. Likewise, user preferences (e.g., the choice of normal forms) can lead to dramatic changes in table structures when expressing conceptually identical schemas. Envisioning the general difficulty for text-to-SQL models to preserve prediction consistency against linguistic and schema variations, we propose MT-Teql, a Metamorphic Testing-based framework for systematically evaluating and augmenting the consistency of TExt-to-SQL models. Inspired by the principles of software metamorphic testing, MT-Teql delivers a model-agnostic framework which implements a comprehensive set of metamorphic relations (MRs) to conduct semantics-preserving transformations toward utterances and schemas. Model Inconsistency can be exposed when the original and transformed inputs induce different SQL queries. In addition, we leverage the transformed inputs to retrain models for further model robustness boost. Our experiments show that our framework exposes thousands of prediction errors from SOTA models and enriches existing datasets by order of magnitude, eliminating over 40% inconsistency errors without compromising standard accuracy.