Text-to-SQL in the Wild: A Naturally-Occurring Dataset Based on Stack Exchange Data

TL;DR

This paper introduces SEDE, a real-world dataset of natural language questions and SQL queries from Stack Exchange, highlighting its diversity and challenges for semantic parsing, and demonstrating the gap in current models' performance.

Contribution

The paper presents SEDE, a naturally-occurring dataset for Text-to-SQL tasks, and proposes an evaluation metric suited for real-world queries, addressing limitations of existing datasets.

Findings

SEDE contains 12,023 real user utterance-SQL pairs.

Current models perform significantly worse on SEDE than on traditional datasets.

The dataset reveals real-world challenges not captured in synthetic datasets.

Abstract

Most available semantic parsing datasets, comprising of pairs of natural utterances and logical forms, were collected solely for the purpose of training and evaluation of natural language understanding systems. As a result, they do not contain any of the richness and variety of natural-occurring utterances, where humans ask about data they need or are curious about. In this work, we release SEDE, a dataset with 12,023 pairs of utterances and SQL queries collected from real usage on the Stack Exchange website. We show that these pairs contain a variety of real-world challenges which were rarely reflected so far in any other semantic parsing dataset, propose an evaluation metric based on comparison of partial query clauses that is more suitable for real-world queries, and conduct experiments with strong baselines, showing a large gap between the performance on SEDE compared to other common datasets.