ST-Raptor: LLM-Powered Semi-Structured Table Question Answering

2025/08/25 by Boyu Niu, Tang, Zirui, Niu, Boyu +13 · 4 citations
Computer Science · Decision Sciences · #Artificial Intelligence (cs.AI) #Data Quality and Management #Databases (cs.DB) #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Topic Modeling #Web Data Mining and Analysis

paper · pdf · doi:10.48550/arxiv.2508.18190

openalex publication_date 2025/08/25 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Semi-structured tables, widely used in real-world applications (e.g., financial reports, medical records, transactional orders), often involve flexible and complex layouts (e.g., hierarchical headers and merged cells). These tables generally rely on human analysts to interpret table layouts and answer relevant natural language questions, which is costly and inefficient. To automate the procedure, existing methods face significant challenges. First, methods like NL2SQL require converting semi-structured tables into structured ones, which often causes substantial information loss. Second, methods like NL2Code and multi-modal LLM QA struggle to understand the complex layouts of semi-structured tables and cannot accurately answer corresponding questions. To this end, we propose ST-Raptor, a tree-based framework for semi-structured table question answering using large language models. First, we introduce the Hierarchical Orthogonal Tree (HO-Tree), a structural model that captures complex semi-structured table layouts, along with an effective algorithm for constructing the tree. Second, we define a set of basic tree operations to guide LLMs in executing common QA tasks. Given a user question, ST-Raptor decomposes it into simpler sub-questions, generates corresponding tree operation pipelines, and conducts operation-table alignment for accurate pipeline execution. Third, we incorporate a two-stage verification mechanism: forward validation checks the correctness of execution steps, while backward validation evaluates answer reliability by reconstructing queries from predicted answers. To benchmark the performance, we present SSTQA, a dataset of 764 questions over 102 real-world semi-structured tables. Experiments show that ST-Raptor outperforms nine baselines by up to 20% in answer accuracy. The code is available at https://github.com/weAIDB/ST-Raptor.

Citations

• LLM/Agent-as-Data-Analyst: A Survey
• Query and Conquer: Execution-Guided SQL Generation
• OpenSearch-SQL: Enhancing Text-to-SQL with Dynamic Few-shot and Consistency Alignment
• Reliable Text-to-SQL with Adaptive Abstention
• TWIX: Automatically Reconstructing Structured Data from Templatized Documents
• Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling
• A Preview of XiYan-SQL: A Multi-Generator Ensemble Framework for Text-to-SQL
• Before Generation, Align it! A Novel and Effective Strategy for Mitigating Hallucinations in Text-to-SQL Generation
• TableLLM: Enabling Tabular Data Manipulation by LLMs in Real Office Usage Scenarios
• mPLUG-DocOwl 1.5: Unified Structure Learning for OCR-free Document Understanding
• PET-SQL: A Prompt-Enhanced Two-Round Refinement of Text-to-SQL with Cross-consistency
• Tables as Texts or Images: Evaluating the Table Reasoning Ability of LLMs and MLLMs
• Multilingual E5 Text Embeddings: A Technical Report
• TAT-LLM: A Specialized Language Model for Discrete Reasoning over Tabular and Textual Data
• TableLlama: Towards Open Large Generalist Models for Tables
• TempTabQA: Temporal Question Answering for Semi-Structured Tables
• SEMQA: Semi-Extractive Multi-Source Question Answering
• Table-GPT: Table-tuned GPT for Diverse Table Tasks
• ReAcTable: Enhancing ReAct for Table Question Answering
• Lost in the Middle: How Language Models Use Long Contexts
• Language Models Enable Simple Systems for Generating Structured Views of Heterogeneous Data Lakes
• Grab What You Need: Rethinking Complex Table Structure Recognition with Flexible Components Deliberation
• GPT-4 Technical Report
• Large Language Models are Versatile Decomposers: Decompose Evidence and Questions for Table-based Reasoning
• Large Language Models Meet NL2Code: A Survey
• Decomposed Prompting: A Modular Approach for Solving Complex Tasks
• PICARD: Parsing Incrementally for Constrained Auto-Regressive Decoding\n from Language Models
• Compositional Semantic Parsing on Semi-Structured Tables
• A relational model of data for large shared data banks

Cited by

• Evolving from Lessons: Skill-Augmented Table Graph Reasoning for Operation-wise Table Question Answering
• A Survey of Data Agents: Emerging Paradigm or Overstated Hype?
• Table Question Answering in the Era of Large Language Models: A Comprehensive Survey of Tasks, Methods, and Evaluation
• LLM/Agent-as-Data-Analyst: A Survey

Related

• Team Anotheroption at SemEval-2025 Task 8: Bridging the Gap Between Open-Source and Proprietary LLMs in Table QA
• Interpretable LLM-based Table Question Answering
• TQA-Bench: Evaluating LLMs for Multi-Table Question Answering
• Same Content, Different Representations: A Controlled Study for Table QA
• TableBench: A Comprehensive and Complex Benchmark for Table Question Answering