SchemaGraphSQL: Efficient Schema Linking with Pathfinding Graph Algorithms for Text-to-SQL on Large-Scale Databases

2025-05-23

AI Summary

To address inefficient schema linking and inaccurate multi-hop JOIN reasoning in large-scale Text-to-SQL, this paper proposes a zero-shot, training-free, graph-path-driven schema linking paradigm. Methodologically, it constructs a foreign-key-constrained schema graph; leverages Gemini 2.5 Flash with lightweight prompts for zero-shot table node extraction; and automatically identifies optimal multi-hop JOIN paths via Dijkstra/A* search, followed by rule-based post-processing to generate executable JOIN sequences. The core contribution is the first zero-shot, fine-tuning-free approach enabling precise multi-hop linkage inference without complex LLM chaining. Evaluated on the BIRD benchmark, it achieves state-of-the-art execution accuracy—outperforming both fine-tuned models and multi-step LLM methods. Moreover, it scales linearly to数千-table schemas while reducing inference cost by over 60%.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• 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
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Text-to-SQL systems translate natural language questions into executable SQL queries, and recent progress with large language models (LLMs) has driven substantial improvements in this task. Schema linking remains a critical component in Text-to-SQL systems, reducing prompt size for models with narrow context windows and sharpening model focus even when the entire schema fits. We present a zero-shot, training-free schema linking approach that first constructs a schema graph based on foreign key relations, then uses a single prompt to Gemini 2.5 Flash to extract source and destination tables from the user query, followed by applying classical path-finding algorithms and post-processing to identify the optimal sequence of tables and columns that should be joined, enabling the LLM to generate more accurate SQL queries. Despite being simple, cost-effective, and highly scalable, our method achieves state-of-the-art results on the BIRD benchmark, outperforming previous specialized, fine-tuned, and complex multi-step LLM-based approaches. We conduct detailed ablation studies to examine the precision-recall trade-off in our framework. Additionally, we evaluate the execution accuracy of our schema filtering method compared to other approaches across various model sizes.

Problem

• Improves schema linking in Text-to-SQL using pathfinding algorithms
• Reduces prompt size for LLMs with narrow context windows
• Enhances SQL query accuracy on large-scale databases

Innovation

• Uses schema graph with foreign key relations
• Applies pathfinding algorithms for optimal joins
• Leverages Gemini 2.5 Flash for table extraction