Orthogonal Hierarchical Decomposition for Structure-Aware Table Understanding with Large Language Models

Bin Cao, Huixian Lu, Chenwen Ma, Ting Wang, Ruizhe Li, Jing Fan
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:11418-11432, 2026.

Abstract

Complex tables with multi-level headers, merged cells and heterogeneous layouts pose persistent challenges for large language models (LLMs) in both understanding and reasoning. Existing approaches typically rely on table linearization or normalized grid modeling. However, these representations struggle to explicitly capture hierarchical structures and cross-dimensional dependencies, which can lead to misalignment between structural semantics and textual representations for non-standard tables. To address this issue, we propose an Orthogonal Hierarchical Decomposition (OHD) framework that constructs structure-preserving input representations of complex tables for LLMs. OHD introduces an Orthogonal Tree Induction (OTI) method based on spatial–semantic co-constraints, which decomposes irregular tables into a column tree and a row tree to capture vertical and horizontal hierarchical dependencies, respectively. Building on this representation, we design a dual-pathway association protocol to symmetrically reconstruct the semantic lineage of each cell, and incorporate an LLM as a semantic arbitrator to align multi-level semantic information. We evaluate OHD framework on two complex table question answering benchmarks, AITQA and HiTab. Experimental results show that OHD consistently outperforms existing representation paradigms across multiple evaluation metrics.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-cao26p, title = {Orthogonal Hierarchical Decomposition for Structure-Aware Table Understanding with Large Language Models}, author = {Cao, Bin and Lu, Huixian and Ma, Chenwen and Wang, Ting and Li, Ruizhe and Fan, Jing}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {11418--11432}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/cao26p/cao26p.pdf}, url = {https://proceedings.mlr.press/v306/cao26p.html}, abstract = {Complex tables with multi-level headers, merged cells and heterogeneous layouts pose persistent challenges for large language models (LLMs) in both understanding and reasoning. Existing approaches typically rely on table linearization or normalized grid modeling. However, these representations struggle to explicitly capture hierarchical structures and cross-dimensional dependencies, which can lead to misalignment between structural semantics and textual representations for non-standard tables. To address this issue, we propose an Orthogonal Hierarchical Decomposition (OHD) framework that constructs structure-preserving input representations of complex tables for LLMs. OHD introduces an Orthogonal Tree Induction (OTI) method based on spatial–semantic co-constraints, which decomposes irregular tables into a column tree and a row tree to capture vertical and horizontal hierarchical dependencies, respectively. Building on this representation, we design a dual-pathway association protocol to symmetrically reconstruct the semantic lineage of each cell, and incorporate an LLM as a semantic arbitrator to align multi-level semantic information. We evaluate OHD framework on two complex table question answering benchmarks, AITQA and HiTab. Experimental results show that OHD consistently outperforms existing representation paradigms across multiple evaluation metrics.} }
Endnote
%0 Conference Paper %T Orthogonal Hierarchical Decomposition for Structure-Aware Table Understanding with Large Language Models %A Bin Cao %A Huixian Lu %A Chenwen Ma %A Ting Wang %A Ruizhe Li %A Jing Fan %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-cao26p %I PMLR %P 11418--11432 %U https://proceedings.mlr.press/v306/cao26p.html %V 306 %X Complex tables with multi-level headers, merged cells and heterogeneous layouts pose persistent challenges for large language models (LLMs) in both understanding and reasoning. Existing approaches typically rely on table linearization or normalized grid modeling. However, these representations struggle to explicitly capture hierarchical structures and cross-dimensional dependencies, which can lead to misalignment between structural semantics and textual representations for non-standard tables. To address this issue, we propose an Orthogonal Hierarchical Decomposition (OHD) framework that constructs structure-preserving input representations of complex tables for LLMs. OHD introduces an Orthogonal Tree Induction (OTI) method based on spatial–semantic co-constraints, which decomposes irregular tables into a column tree and a row tree to capture vertical and horizontal hierarchical dependencies, respectively. Building on this representation, we design a dual-pathway association protocol to symmetrically reconstruct the semantic lineage of each cell, and incorporate an LLM as a semantic arbitrator to align multi-level semantic information. We evaluate OHD framework on two complex table question answering benchmarks, AITQA and HiTab. Experimental results show that OHD consistently outperforms existing representation paradigms across multiple evaluation metrics.
APA
Cao, B., Lu, H., Ma, C., Wang, T., Li, R. & Fan, J.. (2026). Orthogonal Hierarchical Decomposition for Structure-Aware Table Understanding with Large Language Models. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:11418-11432 Available from https://proceedings.mlr.press/v306/cao26p.html.

Related Material