Segment-driven Structural Induction and Semantic Alignment for Heterogeneous Tabular Representation

Woojun Jung, Susik Yoon
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:55090-55109, 2026.

Abstract

Real-world domains often contain heterogeneous tables whose headers vary while their underlying attribute semantics are shared, making it difficult to induce domain-specialized semantics from table-local evidence alone. Existing encoders model parts of this problem, but often underuse column-level value distributions and apply uniform objectives across attributes with different semantic roles. We propose NAVI, a segment-centric pretraining framework that treats each header–value pair as the unit for aggregating schema-level structural evidence and column-level distributional evidence. We realize this design through Masked Segment Modeling and Entropy-driven Segment Alignment, which jointly enforce structured header–value coupling and semantic alignment across stable and instance-specific attributes. Experiments on heterogeneous in-domain tables show improved reconstruction, semantic consistency, and downstream utility across evaluation settings overall.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-jung26i, title = {Segment-driven Structural Induction and Semantic Alignment for Heterogeneous Tabular Representation}, author = {Jung, Woojun and Yoon, Susik}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {55090--55109}, 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/jung26i/jung26i.pdf}, url = {https://proceedings.mlr.press/v306/jung26i.html}, abstract = {Real-world domains often contain heterogeneous tables whose headers vary while their underlying attribute semantics are shared, making it difficult to induce domain-specialized semantics from table-local evidence alone. Existing encoders model parts of this problem, but often underuse column-level value distributions and apply uniform objectives across attributes with different semantic roles. We propose NAVI, a segment-centric pretraining framework that treats each header–value pair as the unit for aggregating schema-level structural evidence and column-level distributional evidence. We realize this design through Masked Segment Modeling and Entropy-driven Segment Alignment, which jointly enforce structured header–value coupling and semantic alignment across stable and instance-specific attributes. Experiments on heterogeneous in-domain tables show improved reconstruction, semantic consistency, and downstream utility across evaluation settings overall.} }
Endnote
%0 Conference Paper %T Segment-driven Structural Induction and Semantic Alignment for Heterogeneous Tabular Representation %A Woojun Jung %A Susik Yoon %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-jung26i %I PMLR %P 55090--55109 %U https://proceedings.mlr.press/v306/jung26i.html %V 306 %X Real-world domains often contain heterogeneous tables whose headers vary while their underlying attribute semantics are shared, making it difficult to induce domain-specialized semantics from table-local evidence alone. Existing encoders model parts of this problem, but often underuse column-level value distributions and apply uniform objectives across attributes with different semantic roles. We propose NAVI, a segment-centric pretraining framework that treats each header–value pair as the unit for aggregating schema-level structural evidence and column-level distributional evidence. We realize this design through Masked Segment Modeling and Entropy-driven Segment Alignment, which jointly enforce structured header–value coupling and semantic alignment across stable and instance-specific attributes. Experiments on heterogeneous in-domain tables show improved reconstruction, semantic consistency, and downstream utility across evaluation settings overall.
APA
Jung, W. & Yoon, S.. (2026). Segment-driven Structural Induction and Semantic Alignment for Heterogeneous Tabular Representation. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:55090-55109 Available from https://proceedings.mlr.press/v306/jung26i.html.

Related Material