Understanding the Effects of Noise in Text-to-SQL: An Examination of the BIRD-Bench Benchmark

TL;DR

An in-depth analysis of the distribution and types of noise in the widely used BIRD-Bench benchmark and the impact of noise on models concludes that informative noise labels and reliable benchmarks are crucial to developing new Text-to-SQL methods that can handle varying types of noise.

Abstract

Text-to-SQL, which involves translating natural language into Structured Query Language (SQL), is crucial for enabling broad access to structured databases without expert knowledge. However, designing models for such tasks is challenging due to numerous factors, including the presence of 'noise,' such as ambiguous questions and syntactical errors. This study provides an in-depth analysis of the distribution and types of noise in the widely used BIRD-Bench benchmark and the impact of noise on models. While BIRD-Bench was created to model dirty and noisy database values, it was not created to contain noise and errors in the questions and gold queries. We found that noise in questions and gold queries are prevalent in the dataset, with varying amounts across domains, and with an uneven distribution between noise types. The presence of incorrect gold SQL queries, which then generate incorrect gold answers, has a significant impact on the benchmark's reliability. Surprisingly, when evaluating models on corrected SQL queries, zero-shot baselines surpassed the performance of state-of-the-art prompting methods. We conclude that informative noise labels and reliable benchmarks are crucial to developing new Text-to-SQL methods that can handle varying types of noise. All datasets, annotations, and code are available at https://github.com/niklaswretblad/the-effects-of-noise-in-text-to-SQL.

Figures (9)

• Figure 1: Example of an incorrect SQL query that generates the wrong gold reference answer for the given question. The JOIN operation incorrectly matches clients and accounts by district_id. Due to the possibility of multiple clients and accounts in the same district, accounts are incorrectly associated with the wrong users.
• Figure 2: Accuracy of various models on Bird-Bench's financial domain. Models are evaluated on the original data (left), corrected SQL queries (middle), and corrected SQL queries and corrected noisy questions.
• Figure 3: Database schema of the database in the financial domain of BIRD-Bench.
• Figure 4: Distribution of question difficulties and execution accuracy of the DIN-SQL model on the different domains of the BIRD-Bench development set.
• Figure 5: Question with ID 125 from the development set of BIRD-Bench which contains syntactical errors and a corrected version of the question.