# M3DocDep: Multi-modal, Multi-page, Multi-document Dependency Chunking with Large Vision-Language Models

arXiv: https://arxiv.org/abs/2605.18774 · HTML: https://mlsanity.com/paper/2605.18774/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Document understanding and multimodal RAG, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 25 of its 48 references that are in the map.
- First posted on arXiv 17 Apr 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Document understanding and multimodal RAG](https://mlsanity.com/mllm/document-understanding-and-multimodal-rag/)
- Retrieval-augmented generation / [Multimodal RAG](https://mlsanity.com/rag/multimodal-rag/)

## Builds on (most cited first, 25 of 25)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Mask R-CNN](https://mlsanity.com/paper/1703.06870/) (2017)
- [Deformable Convolutional Networks](https://mlsanity.com/paper/1703.06211/) (2017)
- [Large Language Models](https://mlsanity.com/paper/2307.05782/) (2023)
- [Deep Biaffine Attention for Neural Dependency Parsing](https://mlsanity.com/paper/1611.01734/) (2016)
- [Scene Text Visual Question Answering](https://mlsanity.com/paper/1905.13648/) (2019)
- [CUAD: An Expert-Annotated NLP Dataset for Legal Contract Review](https://mlsanity.com/paper/2103.06268/) (2021)
- [DocLayNet: A Large Human-Annotated Dataset for Document-Layout Analysis](https://mlsanity.com/paper/2206.01062/) (2022)
- [Document Understanding Dataset and Evaluation (DUDE)](https://mlsanity.com/paper/2305.08455/) (2023)
- [DocLLM: A layout-aware generative language model for multimodal document understanding](https://mlsanity.com/paper/2401.00908/) (2023)
- [Hierarchical multimodal transformers for Multi-Page DocVQA](https://mlsanity.com/paper/2212.05935/) (2022)
- [Can LLM Graph Reasoning Generalize beyond Pattern Memorization?](https://mlsanity.com/paper/2406.15992/) (2024)
- [Financial Report Chunking for Effective Retrieval Augmented Generation](https://mlsanity.com/paper/2402.05131/) (2024)
- [Is Semantic Chunking Worth the Computational Cost?](https://mlsanity.com/paper/2410.13070/) (2024)
- [LumberChunker: Long-Form Narrative Document Segmentation](https://mlsanity.com/paper/2406.17526/) (2024)
- [Recurrent Chunking Mechanisms for Long-Text Machine Reading Comprehension](https://mlsanity.com/paper/2005.08056/) (2020)
- [A Study on the Implementation of Generative AI Services Using an Enterprise Data-Based LLM Application Architecture](https://mlsanity.com/paper/2309.01105/) (2023)
- [LayoutLLM: Large Language Model Instruction Tuning for Visually Rich Document Understanding](https://mlsanity.com/paper/2403.14252/) (2024)
- [HRDoc: Dataset and Baseline Method Toward Hierarchical Reconstruction of Document Structures](https://mlsanity.com/paper/2303.13839/) (2023)
- [Revisiting the tree edit distance and its backtracing: A tutorial](https://mlsanity.com/paper/1805.06869/) (2018)
- [Multi-Page Document Visual Question Answering using Self-Attention Scoring Mechanism](https://mlsanity.com/paper/2404.19024/) (2024)
- [S2 Chunking: A Hybrid Framework for Document Segmentation Through Integrated Spatial and Semantic Analysis](https://mlsanity.com/paper/2501.05485/) (2025)
- [DSG: An End-to-End Document Structure Generator](https://mlsanity.com/paper/2310.09118/) (2023)
- [MultiDocFusion: Hierarchical and Multimodal Chunking Pipeline for Enhanced RAG on Long Industrial Documents](https://mlsanity.com/paper/2604.12352/) (2026)

## Abstract (arXiv)

In long, multi-page industrial documents, retrieval-augmented generation (RAG) depends heavily on whether chunk boundaries follow the document's true structure. Existing text-centric chunkers and generative hierarchy parsers often miss cross-page parent-child relations, figure/table-caption bindings, and boundary cues, which leads to fragmented or redundant chunks and degrades both retrieval and answer quality. We propose M3DocDep, an LVLM-based pipeline that first recovers block-level dependencies and then constructs chunks along the recovered document tree. The pipeline uses SharedDet as a common DP+OCR preprocessing layer, extracts multimodal block embeddings with boundary-aware SoftROI pooling, scores candidate parent-child edges with a biaffine head, decodes a globally valid dependency tree with MST constraints, and builds tree-guided chunks annotated with section paths and page ranges. Under a shared-block evaluation protocol, M3DocDep improves STEDS by +28.5 to +39.6 percent on DHP benchmarks, retrieval nDCG by +1.1 to +15.3 percent, and QA ANLS by +4.5 to +15.3 percent on corpus-level RAG benchmarks. These results show that recovering document dependencies before chunking yields more coherent retrieval units for long, multi-page multimodal documents.