JoinGR: Learning to Traverse Join Graphs for Table Retrieval

2026-10-01

AI Summary

This study addresses the limitation of conventional dense retrieval methods that overlook inter-table join relationships, rendering them ineffective at recalling tables requiring multi-hop associations for identification. To overcome this, we propose a join-aware table retrieval framework that treats the database join graph as the retrieval space, aggregating scores through anchor table selection and join edge traversal. Furthermore, a query condition scorer is designed to efficiently capture reusable graph traversal patterns via lightweight training, combining a multilayer perceptron with pairwise margin loss over frozen pretrained embeddings. The proposed method achieves strong performance on the BIRD and Spider benchmarks and significantly improves multi-hop table recall on the BEAVER enterprise benchmark, while demonstrating robust cross-domain transferability.

Abstract

Retrieving the right tables is a prerequisite for Text-to-SQL over realistic databases. Dense table retrievers rank schema elements independently, but this ignores a key source of evidence: some required tables are not mentioned in the question and become identifiable only through their join relationships to already relevant tables. We introduce JOINGR, a join-aware table retrieval method that treats the database join graph as the retrieval space. Columns are represented as graph nodes, while intra-table and foreign-key relationships are represented as typed edges. Given a question, JOINGR selects semantically similar anchor tables, traverses join edges with a query-conditioned scorer, and aggregates the resulting edge deposits into table scores. The scorer is a lightweight MLP on top of frozen query, node, and edge embeddings, trained with a pairwise margin loss over gold tables. On BIRD and Spider datasets, JOINGR is competitive with the strongest retrieval baselines. On BEAVER, a challenging enterprise benchmark with multi-hop table requirements, JOINGR substantially improves recall over dense retrieval and re-ranking baselines. Cross-domain experiments show that the learned scorer transfers across benchmarks, indicating that the method captures reusable joingraph traversal behavior.

Problem

• Table Retrieval
• Text-to-SQL
• Join Graph
• Dense Retrieval

Innovation

• Table Retrieval
• Join Graph Traversal
• Text-to-SQL
• Graph-based Scorer
• Cross-domain Transfer