Guided Table Retrieval for Structured Data Search

2026-08-12

AI Summary

This work addresses the challenge of accurately identifying relevant tables and their join relationships in natural language to structured database question answering by integrating schema knowledge with user intent. The authors propose a four-stage guided table retrieval method: first, a hash-based predictor identifies candidate tables; second, reachability is explored via a join graph; third, large language models resolve source/target semantic ambiguities; and finally, a minimal topologically ordered join tree is algorithmically constructed. By decoupling the task into deterministic localization, coverage-aware exploration, semantic reasoning, and structural coherence, the approach strategically leverages large language models at critical stages to enhance robustness. Evaluated on the BIRD-DEV and BEAVER benchmarks, the method achieves exact-match accuracies of 94% and 70% (with F1 scores of 92% and 53%, respectively), substantially outperforming existing approaches while producing precise join structures directly usable for query compilation.

Abstract

Answering natural language questions over structured databases requires identifying the relevant tables and determining how to join them---a task that demands both schema knowledge and semantic understanding of the user's intent.We present guided table retrieval, a four-phase pipeline that combines deterministic grounding via hash-based predictors, structural exploration of join-graph reachability, LLM-powered disambiguation of sources and targets, and algorithmic merging into minimal, topologically ordered join trees. By decomposing the problem into phases with distinct responsibilities--- determinism, coverage, semantic reasoning, and coherence--- the pipeline avoids the brittleness of end-to-end LLM approaches while leveraging LLMs where their contextual judgment is most needed. We evaluate on BIRD-DEV and the enterprise-scale BEAVER benchmark, achieving 94% and 70% precision respectively, with 92% and 53% F1---substantially outperforming existing baselines on precision and F1 while producing exact join trees that can be directly consumed by downstream query compilers.

Problem

• table retrieval
• structured data search
• schema understanding
• join discovery
• natural language interfaces

Innovation

• guided table retrieval
• join tree construction
• structured data search
• LLM-powered disambiguation
• deterministic grounding