Text-to-SQL as a long-standing open problem

Develop automatic Text-to-SQL systems that translate natural-language queries into executable SQL for a specified relational database schema to improve database accessibility without requiring SQL expertise.

Background

The paper notes that Text-to-SQL aims to automate the generation of SQL queries from natural language to make databases accessible to users without SQL expertise. Although orthogonal to the paper’s focus on SQL-to-SQL rewriting, the authors explicitly characterize Text-to-SQL as a long-standing open problem, referencing early work in database question answering.

This acknowledgment situates Text-to-SQL as an enduring challenge in the broader landscape of natural language interfaces to databases, underscoring the difficulty of reliably mapping language to precise SQL semantics across schemas and query types.