Automatic database description generation for Text-to-SQL

2025-02-28

AI Summary

To bridge the semantic gap in Text-to-SQL caused by the absence of natural language descriptions for database schemas, this paper proposes a dual-path collaborative schema description generation method. We introduce, for the first time, a bidirectional generation paradigm—“coarse-to-fine” and “fine-to-coarse”—to jointly capture global structural understanding and local semantic precision. Our approach employs a two-stage, large language model–based prompting framework that integrates schema structure analysis with hierarchical semantic reasoning. Evaluated on the Bird benchmark, our method improves SQL execution accuracy by 0.93% and achieves description quality at 37% of human-written level, significantly narrowing the semantic gap between natural language and SQL. The source code is publicly available.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Planning, Routing, and Scheduling: Planning with Language Models
• Knowledge Representation and Reasoning: Description Logics

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

In the context of the Text-to-SQL task, table and column descriptions are crucial for bridging the gap between natural language and database schema. This report proposes a method for automatically generating effective database descriptions when explicit descriptions are unavailable. The proposed method employs a dual-process approach: a coarse-to-fine process, followed by a fine-to-coarse process. The coarse-to-fine approach leverages the inherent knowledge of LLM to guide the understanding process from databases to tables and finally to columns. This approach provides a holistic understanding of the database structure and ensures contextual alignment. Conversely, the fine-to-coarse approach starts at the column level, offering a more accurate and nuanced understanding when stepping back to the table level. Experimental results on the Bird benchmark indicate that using descriptions generated by the proposed improves SQL generation accuracy by 0.93% compared to not using descriptions, and achieves 37% of human-level performance. The source code is publicly available at https://github.com/XGenerationLab/XiYan-DBDescGen.

Problem

• Generates database descriptions for Text-to-SQL tasks.
• Uses dual-process approach for accurate schema understanding.
• Improves SQL generation accuracy by 0.93%.

Innovation

• Dual-process approach for database description generation
• Coarse-to-fine and fine-to-coarse understanding processes
• Improves SQL generation accuracy by 0.93%