CRED-SQL: Enhancing Real-world Large Scale Database Text-to-SQL Parsing through Cluster Retrieval and Execution Description

2025-08-18

AI Summary

To address semantic misalignment between natural language questions (NLQs) and SQL in large-scale databases—causing challenges in schema linking and generation drift—this paper proposes CRED-SQL, a novel Text-to-SQL framework. Methodologically, it introduces Executable Description Language (EDL) as an interpretable, executable intermediate representation, decomposing Text-to-SQL into two sequential stages: Text-to-EDL and EDL-to-SQL. It further designs a clustering-driven schema retrieval mechanism to enhance semantic alignment between NLQs and schema elements, and integrates LLM-powered NL rewriting with EDL generation. Evaluated on SpiderUnion and BirdUnion benchmarks, CRED-SQL achieves new state-of-the-art performance, demonstrating significant improvements in cross-domain generalization, execution accuracy, and system scalability.

Technology Category

• Natural Language Processing: Information Extraction
• Data Mining & Knowledge Management: Intelligent Query Processing
• Knowledge Representation and Reasoning: Description Logics

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web 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

Recent advances in large language models (LLMs) have significantly improved the accuracy of Text-to-SQL systems. However, a critical challenge remains: the semantic mismatch between natural language questions (NLQs) and their corresponding SQL queries. This issue is exacerbated in large-scale databases, where semantically similar attributes hinder schema linking and semantic drift during SQL generation, ultimately reducing model accuracy. To address these challenges, we introduce CRED-SQL, a framework designed for large-scale databases that integrates Cluster Retrieval and Execution Description. CRED-SQL first performs cluster-based large-scale schema retrieval to pinpoint the tables and columns most relevant to a given NLQ, alleviating schema mismatch. It then introduces an intermediate natural language representation-Execution Description Language (EDL)-to bridge the gap between NLQs and SQL. This reformulation decomposes the task into two stages: Text-to-EDL and EDL-to-SQL, leveraging LLMs' strong general reasoning capabilities while reducing semantic deviation. Extensive experiments on two large-scale, cross-domain benchmarks-SpiderUnion and BirdUnion-demonstrate that CRED-SQL achieves new state-of-the-art (SOTA) performance, validating its effectiveness and scalability. Our code is available at https://github.com/smduan/CRED-SQL.git

Problem

• Addresses semantic mismatch in Text-to-SQL for large databases
• Improves schema linking via cluster retrieval for accurate SQL generation
• Introduces Execution Description Language to reduce semantic deviation

Innovation

• Cluster-based schema retrieval for relevant tables
• Execution Description Language bridges NLQ-SQL gap
• Two-stage Text-to-EDL and EDL-to-SQL decomposition