DeepTable: Structural Attention Biases and Tree Path Encoding for Hierarchical Table Understanding

• URL: http://arxiv.org/abs/2609.07707v1
• Date: Mon, 07 Sep 2026 16:19:33 GMT
• Title: DeepTable: Structural Attention Biases and Tree Path Encoding for Hierarchical Table Understanding
• Abstract summary: Large language models (LLMs) have demonstrated strong performance in table understanding.<n>We propose DeepTable, a structure-aware approach for table understanding with LLMs.
• Score: 10.761232681902394
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) have demonstrated strong performance in table understanding. However, they typically process table content and headers as linearized token sequences. This representation weakens the two-dimensional and hierarchical structural relationships encoded by multi-level row and column headers. Existing parameter-efficient fine-tuning methods incorporate basic row and column information but do not explicitly capture the rich structural dependencies induced by hierarchical table headers. We propose DeepTable, a structure-aware approach for table understanding with LLMs. DeepTable comprises two complementary components. Structural Attention Bias (SAB) introduces learnable biases into the attention logits to explicitly represent whether pairs of table tokens share the same row or column. Tree Path Encoding (TPE) represents each table token using the ancestor paths of its row and column headers, preserving its position within the multi-level table structure. We integrate DeepTable with TableLoRA (He et al., 2025) to inject structural information into parameter-efficient adaptation. Across three LLM backbones, DeepTable consistently improves the corresponding TableLoRA baselines on three table question answering benchmarks, achieving average gains of 7.42 points on HiTab, 3.23 points on WikiTQ, and 2.01 BLEU points on FeTaQA. These results demonstrate the effectiveness of the proposed structural biases across different LLM backbones.

Related papers

• H2Table: Hierarchical Hypergraph-Enhanced Large Language Models for Complex Table Reasoning [16.150378428877445]
  H2Table is a novel framework that represents complex tables as hierarchical nested hypergraphs.<n>We design a tailored hypergraph encoder to facilitate message passing between hyperedges (headers) and nodes (cells)
  arXiv  Detail & Related papers  (2026-09-01T13:19:42Z)
• TabEmb: Joint Semantic-Structure Embedding for Table Annotation [3.03279900330493]
  Table annotation is crucial for making web and enterprise tables usable in downstream NLP applications.<n>Existing models learn by linearizing the 2D table into a 1D token sequence and encoding it with pretrained language models.<n>We propose TabEmb, which directly targets these limitations by decoupling semantic encoding from structural modeling.
  arXiv  Detail & Related papers  (2026-04-21T00:25:49Z)
• TDATR: Improving End-to-End Table Recognition via Table Detail-Aware Learning and Cell-Level Visual Alignment [70.83664203825235]
  TDATR (Table Detail-Aware Table Recognition) improves end-to-end TR through table detail-aware learning and cell-level visual alignment.<n>We achieve state-of-the-art or highly competitive performance on seven benchmarks without dataset-specific fine-tuning.
  arXiv  Detail & Related papers  (2026-03-24T05:45:02Z)
• Orthogonal Hierarchical Decomposition for Structure-Aware Table Understanding with Large Language Models [13.58784346599112]
  We propose an Orthogonal Hierarchical Decomposition (OHD) framework that constructs structure-preserving input representations of complex tables for LLMs.<n>Building on this representation, we design a dual-pathway association protocol to symmetrically reconstruct semantic lineage of each cell.<n>We evaluate OHD framework on two complex table question answering benchmarks, AITQA and HiTab.
  arXiv  Detail & Related papers  (2026-02-02T11:22:43Z)
• TableLoRA: Low-rank Adaptation on Table Structure Understanding for Large Language Models [57.005158277893194]
  TableLoRA is a module designed to improve LLMs' understanding of table structure during PEFT.<n>It incorporates special tokens for serializing tables with special token encoder and uses 2D LoRA to encode low-rank information on cell positions.
  arXiv  Detail & Related papers  (2025-03-06T12:50:14Z)
• TAP4LLM: Table Provider on Sampling, Augmenting, and Packing Semi-structured Data for Large Language Model Reasoning [55.33939289989238]
  We propose TAP4LLM as a versatile pre-processor suite for leveraging large language models (LLMs) in table-based tasks effectively. It covers several distinct components: (1) table sampling to decompose large tables into manageable sub-tables based on query semantics, (2) table augmentation to enhance tables with additional knowledge from external sources or models, and (3) table packing & serialization to convert tables into various formats suitable for LLMs' understanding.
  arXiv  Detail & Related papers  (2023-12-14T15:37:04Z)
• SEMv2: Table Separation Line Detection Based on Instance Segmentation [96.36188168694781]
  We propose an accurate table structure recognizer, termed SEMv2 (SEM: Split, Embed and Merge) We address the table separation line instance-level discrimination problem and introduce a table separation line detection strategy based on conditional convolution. To comprehensively evaluate the SEMv2, we also present a more challenging dataset for table structure recognition, dubbed iFLYTAB.
  arXiv  Detail & Related papers  (2023-03-08T05:15:01Z)
• Table Retrieval May Not Necessitate Table-specific Model Design [83.27735758203089]
  We focus on the task of table retrieval, and ask: "is table-specific model design necessary for table retrieval?" Based on an analysis on a table-based portion of the Natural Questions dataset (NQ-table), we find that structure plays a negligible role in more than 70% of the cases. We then experiment with three modules to explicitly encode table structures, namely auxiliary row/column embeddings, hard attention masks, and soft relation-based attention biases. None of these yielded significant improvements, suggesting that table-specific model design may not be necessary for table retrieval.
  arXiv  Detail & Related papers  (2022-05-19T20:35:23Z)
• TGRNet: A Table Graph Reconstruction Network for Table Structure Recognition [76.06530816349763]
  We propose an end-to-end trainable table graph reconstruction network (TGRNet) for table structure recognition. Specifically, the proposed method has two main branches, a cell detection branch and a cell logical location branch, to jointly predict the spatial location and the logical location of different cells.
  arXiv  Detail & Related papers  (2021-06-20T01:57:05Z)
• TUTA: Tree-based Transformers for Generally Structured Table Pre-training [47.181660558590515]
  Recent attempts on table understanding mainly focus on relational tables, yet overlook to other common table structures. We propose TUTA, a unified pre-training architecture for understanding generally structured tables. TUTA is highly effective, achieving state-of-the-art on five widely-studied datasets.
  arXiv  Detail & Related papers  (2020-10-21T13:22:31Z)
• Identifying Table Structure in Documents using Conditional Generative Adversarial Networks [0.0]
  In many industries and in academic research, information is primarily transmitted in the form of unstructured documents. We propose a top-down approach, first using a conditional generative adversarial network to map a table image into a standardised skeleton' table form. We then deriving latent table structure using xy-cut projection and Genetic Algorithm optimisation.
  arXiv  Detail & Related papers  (2020-01-13T20:42:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.