HI-SQL: Optimizing Text-to-SQL Systems through Dynamic Hint Integration

2025-06-11

IEEE International Joint Conference on Neural Network

AI Summary

Text-to-SQL systems suffer from low accuracy, high latency, and severe error propagation when handling complex queries involving multi-table joins and nested conditions—primarily due to reliance on multi-stage pipelines and manually engineered prompts. To address this, we propose a history-aware dynamic prompt generation mechanism that leverages historical query logs to inject lightweight, context-sensitive hints directly into the input sequence. Our approach tightly integrates query log analysis, adaptive prompt engineering, and end-to-end LLM inference—marking the first such unification. It eliminates multi-stage processing and manual intervention entirely. Evaluated on Spider and BIRD benchmarks, our method improves execution accuracy by +8.2%, reduces LLM API calls by 37%, and cuts end-to-end latency by 29%. The solution achieves both computational efficiency and practical deployability.

Technology Category

• Natural Language Processing: Prompt Engineering / Prompting
• Data Mining & Knowledge Management: Intelligent Query Processing
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Text-to-SQL generation bridges the gap between natural language and databases, enabling users to query data without requiring SQL expertise. While large language models (LLMs) have significantly advanced the field, challenges remain in handling complex queries that involve multi-table joins, nested conditions, and intricate operations. Existing methods often rely on multi-step pipelines that incur high computational costs, increase latency, and are prone to error propagation. To address these limitations, we propose HI-SQL, a pipeline that incorporates a novel hint generation mechanism utilizing historical query logs to guide SQL generation. By analyzing prior queries, our method generates contextual hints that focus on handling the complexities of multi-table and nested operations. These hints are seamlessly integrated into the SQL generation process, eliminating the need for costly multi-step approaches and reducing reliance on human-crafted prompts. Experimental evaluations on multiple benchmark datasets demonstrate that our approach significantly improves query accuracy of LLM-generated queries while ensuring efficiency in terms of LLM calls and latency, offering a robust and practical solution for enhancing Text-to-SQL systems.

Problem

• Optimizes Text-to-SQL systems for complex queries
• Reduces computational costs and latency in SQL generation
• Improves accuracy by integrating dynamic hints from query logs

Innovation

• Dynamic hint generation from historical query logs
• Seamless integration of contextual hints into SQL generation
• Eliminates costly multi-step pipelines for efficiency