English

Structured Attention Matters to Multimodal LLMs in Document Understanding

Computation and Language 2025-06-30 v1 Artificial Intelligence Information Retrieval

Abstract

Document understanding remains a significant challenge for multimodal large language models (MLLMs). While previous research has primarily focused on locating evidence pages through precise multimodal queries, our work investigates a fundamental yet overlooked aspect: how input format influences document comprehension performance. Through systematic analysis, we discover that raw OCR text often impairs rather than improves MLLMs' performance, which is a counterintuitive finding we attribute to attention dispersion and structure loss. To further substantiate our hypothesis, we propose a novel structure-preserving approach that encodes document elements using the LaTex paradigm, maintaining the hierarchical organization and spatial relationships critical for comprehension. Our attention analysis reveals that structured text induces structured attention patterns on both textual and visual content, directing models to focus on semantically meaningful regions while reducing attention waste. This approach significantly enhances MLLMs' document question answering performance across diverse document types without requiring architectural modifications or additional training.

Keywords

Cite

@article{arxiv.2506.21600,
  title  = {Structured Attention Matters to Multimodal LLMs in Document Understanding},
  author = {Chang Liu and Hongkai Chen and Yujun Cai and Hang Wu and Qingwen Ye and Ming-Hsuan Yang and Yiwei Wang},
  journal= {arXiv preprint arXiv:2506.21600},
  year   = {2025}
}
R2 v1 2026-07-01T03:35:06.908Z