Cost-Effective Numerical QA over Semi-Structured Table: Structuring, Resolution, Planning

2026-10-06

AI Summary

This study addresses the challenges of structural comprehension, semantic ambiguity, and high reasoning costs in numerical question answering over semi-structured tables by proposing the SemiBonsai framework. This approach pioneers a holistic methodology that integrates table-to-tree structural conversion with budget constraints, encompassing four core modules: multi-path hierarchical structure parsing, context-aware disambiguation, plan-guided reasoning, and Bandit-based budget-aware routing. Experimental results demonstrate that the proposed framework significantly improves both question answering accuracy and cost-efficiency across three benchmark datasets. Specifically, structure parsing accuracy increases by 46%, and the router achieves an average relative gain of 5%. These findings indicate that SemiBonsai effectively balances model performance with computational economy, offering a practical solution for resource-constrained tabular reasoning tasks.

Abstract

Semi-structured tables encode rich semantic information through diverse layout elements, such as hierarchical row and column headers. Answering numerical questions over such data is challenging because it requires a joint effort of accurate structural understanding of tables, question uncertainty resolution, and complex query intents interpretation. Existing methods are rarely effective in handling the above multifaceted challenges in a holistic manner; moreover, they rely on LLMs without considering financial implications. We propose SemiBonsai, a cost-effective, LLM-powered framework featuring: (1) a Multiway Layered Table Structurer that converts a table into a multiway layered tree organized by structural elements; (2) a Context-Aware Uncertainty Resolver that grounds underspecified phrases to context-coherent structural elements; (3) a Plan-Guided Reasoner that decomposes complex query intents into a query plan for reasoning; (4) a Budget-Aware Bandit Router that learns per-instance LLM utilities and optimizes LLM selection under fixed monetary constraints. Experiments on three benchmarks show that SemiBonsai improves both QA effectiveness and cost-effectiveness. Notably, the Table Structurer improves answer accuracy by 46% over alternative tree construction methods, and Router achieves a 5% average relative gain in answer accuracy compared with existing routing methods across budgets.

Problem

• Semi-structured table
• Numerical question answering
• Structural understanding
• Uncertainty resolution
• Cost-effectiveness

Innovation

• Semi-structured Table QA
• Multiway Layered Tree
• Plan-Guided Reasoning
• Budget-Aware Routing
• Cost-Effective LLM