[2604.08999] ASTRA: Adaptive Semantic Tree Reasoning Architecture for Complex Table Question Answering

Computer Science > Computation and Language arXiv:2604.08999 (cs) [Submitted on 10 Apr 2026] Title:ASTRA: Adaptive Semantic Tree Reasoning Architecture for Complex Table Question Answering Authors:Xiaoke Guo, Songze Li, Zhiqiang Liu, Zhaoyan Gong, Yuanxiang Liu, Huajun Chen, Wen Zhang View a PDF of the paper titled ASTRA: Adaptive Semantic Tree Reasoning Architecture for Complex Table Question Answering, by Xiaoke Guo and 6 other authors View PDF HTML (experimental) Abstract:Table serialization remains a critical bottleneck for Large Language Models (LLMs) in complex table question answering, hindered by challenges such as structural neglect, representation gaps, and reasoning opacity. Existing serialization methods fail to capture explicit hierarchies and lack schema flexibility, while current tree-based approaches suffer from limited semantic adaptability. To address these limitations, we propose ASTRA (Adaptive Semantic Tree Reasoning Architecture) including two main modules, AdaSTR and DuTR. First, we introduce AdaSTR, which leverages the global semantic awareness of LLMs to reconstruct tables into Logical Semantic Trees. This serialization explicitly models hierarchical dependencies and employs an adaptive mechanism to optimize construction strategies based on table scale. Second, building on this structure, we present DuTR, a dual-mode reasoning framework that integrates tree-search-based textual navigation for linguistic alignment and symbolic code execution for pre...