TableReasoner: Advancing Table Reasoning Framework with Large Language Models

2025-07-10

AI Summary

Real-world table question answering (TQA) faces challenges including large-scale tables, incomplete column semantics, and entity ambiguity. Method: This paper proposes a novel framework integrating large language models (LLMs) with programmatic reasoning. It (1) introduces a multi-step schema linking strategy that dynamically generates structure-focused table representations to mitigate semantic ambiguity; (2) designs an iterative “Think–Reason–Reflect” architecture for joint structural and semantic modeling; and (3) incorporates LLM-driven programmable reasoning to generate interpretable, executable SQL-like queries. Contribution/Results: The framework achieves first place on both subtasks of SemEval-2025 Task 8, significantly improving reasoning accuracy and robustness on complex, large-scale, and low-quality tables. It establishes a new paradigm for semantic understanding of real-world tabular data characterized by structural sparsity and lexical ambiguity.

Technology Category

• Natural Language Processing: Question Answering
• Knowledge Representation and Reasoning: Logic Programming
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

The paper presents our system developed for table question answering (TQA). TQA tasks face challenges due to the characteristics of real-world tabular data, such as large size, incomplete column semantics, and entity ambiguity. To address these issues, we propose a large language model (LLM)-powered and programming-based table reasoning framework, named TableReasoner. It models a table using the schema that combines structural and semantic representations, enabling holistic understanding and efficient processing of large tables. We design a multi-step schema linking plan to derive a focused table schema that retains only query-relevant information, eliminating ambiguity and alleviating hallucinations. This focused table schema provides precise and sufficient table details for query refinement and programming. Furthermore, we integrate the reasoning workflow into an iterative thinking architecture, allowing incremental cycles of thinking, reasoning and reflection. Our system achieves first place in both subtasks of SemEval-2025 Task 8.

Problem

• Addressing challenges in table question answering due to large, ambiguous data
• Proposing a framework for holistic table understanding using LLMs
• Enhancing reasoning with multi-step schema linking and iterative workflows

Innovation

• LLM-powered programming-based table reasoning framework
• Multi-step schema linking for focused table schema
• Iterative thinking architecture for reasoning workflow