FGTR: Fine-Grained Multi-Table Retrieval via Hierarchical LLM Reasoning

TL;DR

FGTR introduces a hierarchical reasoning approach for fine-grained multi-table retrieval using LLMs, significantly improving accuracy and efficiency over existing methods.

Contribution

It proposes a novel hierarchical multi-table retrieval paradigm leveraging LLM reasoning, with new benchmarks and superior performance on existing datasets.

Findings

FGTR improves F2 metric by 18% on Spider.

FGTR improves F2 metric by 21% on BIRD.

Hierarchical reasoning enhances retrieval accuracy and efficiency.

Abstract

With the rapid advancement of large language models (LLMs), growing efforts have been made on LLM-based table retrieval. However, existing studies typically focus on single-table query, and implement it by similarity matching after encoding the entire table. These methods usually result in low accuracy due to their coarse-grained encoding which incorporates much query-irrelated data, and are also inefficient when dealing with large tables, failing to fully utilize the reasoning capabilities of LLM. Further, multi-table query is under-explored in retrieval tasks. To this end, we propose a hierarchical multi-table query method based on LLM: Fine-Grained Multi-Table Retrieval FGTR, a new retrieval paradigm that employs a human-like reasoning strategy. Through hierarchical reasoning, FGTR first identifies relevant schema elements and then retrieves the corresponding cell contents, ultimately constructing a concise and accurate sub-table that aligns with the given query. To comprehensively evaluate the performance of FGTR, we construct two new benchmark datasets based on Spider and BIRD . Experimental results show that FGTR outperforms previous state-of-the-art methods, improving the F_2 metric by 18% on Spider and 21% on BIRD, demonstrating its effectiveness in enhancing fine-grained retrieval and its potential to improve end-to-end performance on table-based downstream tasks.