Uncovering Limitations of Large Language Models in Information Seeking from Tables

TL;DR

This paper evaluates large language models' ability to seek information from tables, introducing a new benchmark that reveals their limitations in understanding table structures and robustness, highlighting the need for improved models.

Contribution

It presents a reliable, question-based benchmark for Table Information Seeking and analyzes LLMs' performance, exposing their deficiencies in understanding tables and robustness issues.

Findings

GPT-4-turbo performs marginally well

Most models perform inadequately in TIS tasks

LLMs struggle with table structure comprehension

Abstract

Tables are recognized for their high information density and widespread usage, serving as essential sources of information. Seeking information from tables (TIS) is a crucial capability for Large Language Models (LLMs), serving as the foundation of knowledge-based Q&A systems. However, this field presently suffers from an absence of thorough and reliable evaluation. This paper introduces a more reliable benchmark for Table Information Seeking (TabIS). To avoid the unreliable evaluation caused by text similarity-based metrics, TabIS adopts a single-choice question format (with two options per question) instead of a text generation format. We establish an effective pipeline for generating options, ensuring their difficulty and quality. Experiments conducted on 12 LLMs reveal that while the performance of GPT-4-turbo is marginally satisfactory, both other proprietary and open-source models perform inadequately. Further analysis shows that LLMs exhibit a poor understanding of table structures, and struggle to balance between TIS performance and robustness against pseudo-relevant tables (common in retrieval-augmented systems). These findings uncover the limitations and potential challenges of LLMs in seeking information from tables. We release our data and code to facilitate further research in this field.