Scaling Beyond Context: A Survey of Multimodal Retrieval-Augmented Generation for Document Understanding

ACL 2026 Main Conference

Sensen Gao^1*, Shanshan Zhao^2†, Xu Jiang³, Lunhao Duan^4*, Yong Xien Chng^3*,
Qing-Guo Chen², Weihua Luo², Kaifu Zhang², Jia-Wang Bian¹, Mingming Gong^1,5†

¹MBZUAI   ²Alibaba International Digital Commerce Group   ³Tsinghua University   ⁴Wuhan University   ⁵University of Melbourne

_{* Work done during an internship at Alibaba International Digital Commerce Group. † Corresponding authors.}

---

This repository maintains a curated list of methods, datasets, and benchmarks for Multimodal Retrieval-Augmented Generation (RAG) in Document Understanding, based on our ACL 2026 survey. We will keep updating this list. Feel free to open an issue or PR if we missed any relevant work!

Updates

• Ongoing - We keep this list actively maintained! If you have a recent paper, or notice any relevant work we missed, feel free to contact us or open an issue/PR — we'll include it in a future update.
• 2026.04 - Paper accepted at ACL 2026 Main Conference!
• 2025.10 - Paper available on arXiv.

Table of Contents

• Overview
• Methods
  • Open-Domain Methods
  • Closed-Domain Methods
  • Graph-based Methods
  • Agent-based Methods
• Datasets & Benchmarks
  • Document Understanding Datasets
  • Multimodal RAG Benchmarks
• Citation

Overview

A taxonomy of multimodal RAG for document understanding: retrieval domain (open vs. closed), retrieval modality (image vs. image+text), retrieval granularity (page- vs. element-level), and hybrid graph-/agent-based enhancements.

Methods

Open-Domain Methods

Method	Venue	Modality	Granularity	Training	Paper
DSE	EMNLP 2024	Image	Page	Yes	Link
ColPali	ICLR 2025	Image	Page	Yes	Link
ColQwen2	ICLR 2025	Image	Page	Yes	Link
VisRAG	ICLR 2025	Image	Page	Yes	Link
M3DocRAG	Preprint	Image	Page	No	Link
VisDoMRAG	NAACL 2025	Image+Text	Page	No	Link
GME	CVPR 2025	Image+Text	Page	Yes	Link
ViDoRAG	EMNLP 2025	Image+Text	Page	No	Link
HM-RAG	ACM MM 2025	Image+Text	Page	No	Link
VDocRAG	CVPR 2025	Image	Page	Yes	Link
VRAG-RL	Preprint	Image	Element	Yes	Link
CoRe-MMRAG	ACL 2025	Image+Text	Page	Yes	Link
Light-ColPali	ACL 2025	Image	Page	Yes	Link
MM-R5	Preprint	Image	Page	Yes	Link
SimpleDoc	Preprint	Image+Text	Page	No	Link
DocVQA-RAP	ICIC 2025	Image	Element	No	Link
RL-QR	Preprint	Image	Page	Yes	Link
Patho-AgenticRAG	Preprint	Image	Page	Yes	Link
M2IO-R1	Preprint	Image+Text	Page	Yes	Link
mKG-RAG	Preprint	Image+Text	Element	Yes	Link
DB3Team-RAG	Preprint	Image+Text	Page	Yes	Link
PREMIR	EMNLP 2025	Image+Text	Element	No	Link
CMRAG	Preprint	Image+Text	Page	No	Link
MoLoRAG	EMNLP 2025	Image	Page	Yes	Link
SERVAL	Preprint	Image	Page	No	Link
MetaEmbed	Preprint	Image	Page	Yes	Link
DocPruner	Preprint	Image	Page	Yes	Link
RECON	Preprint	Image+Text	Element	No	Link
LAD-RAG	Preprint	Image+Text	Element	No	Link
HEAVEN	Preprint	Image	Page	No	Link
MARA	ACM MM 2025	Image	Element	Yes	Link
HPC-ColPali	Preprint	Image	Page	Yes	Link
RegionRAG	Preprint	Image	Element	Yes	Link
IndustryRAG	EMNLP Industry 2025	Image	Page	No	Link
COLMATE	EMNLP Industry 2025	Image	Page	Yes	Link
LILaC	EMNLP 2025	Image	Element	No	Link
HKRAG	Preprint	Image	Element	Yes	Link
SLEUTH	Preprint	Image	Page	No	Link
Snappy	Preprint	Image	Element	No	Link

Closed-Domain Methods

Method	Venue	Modality	Granularity	Training	Paper
CREAM	ACM MM 2024	Image+Text	Page	Yes	Link
SV-RAG	ICLR 2025	Image	Page	Yes	Link
FRAG	Preprint	Image	Page	No	Link
MG-RAG	Preprint	Image+Text	Element	No	Link
VisChunk	Preprint	Image+Text	Page	No	Link
MMRAG-DocQA	Preprint	Image+Text	Element	No	Link
ReDocRAG	ICDAR WML 2025	Image	Page	Yes	Link
DREAM	ACM MM 2025	Image	Page	Yes	Link
HEAR	ACM MMW 2025	Image+Text	Page	No	Link

Graph-based Methods

Method	Venue	Key Idea	Paper
HM-RAG	ACM MM 2025	Hierarchical multi-agent framework with graph databases for structured relation capture	Link
mKG-RAG	Preprint	Multimodal knowledge graphs aligning entities across vision and text	Link
DB3Team-RAG	Preprint	Image-indexed knowledge graphs for domain-specific retrieval	Link
MoLoRAG	EMNLP 2025	Page graphs encoding logical connections via graph traversal	Link
RECON	Preprint	Global multimodal document graph linking intra-page and inter-page relations	Link
LAD-RAG	Preprint	Layout-aware component graphs with dynamic traversal	Link
LILaC	EMNLP 2025	Layered component graph with late interaction subgraph retrieval	Link

Agent-based Methods

Method	Venue	Key Idea	Paper
ViDoRAG	EMNLP 2025	Iterative agent workflow with exploration, summarization, and reflection	Link
HM-RAG	ACM MM 2025	Hierarchical multi-agent with query decomposition and consistency voting	Link
Patho-AgenticRAG	Preprint	Task decomposition and multi-turn search for pathology textbooks	Link
HEAR	ACM MMW 2025	Closed-loop multi-agent reasoning with VLM-based document parsing	Link
SLEUTH	Preprint	Coarse-to-fine agent filtering and distilling salient evidence	Link

Datasets & Benchmarks

Document Understanding Datasets

Dataset	#Queries	#Docs/Images	Content	Paper
TabFQuAD	210	210 (I)	Table	Link
PlotQA	28.9M	224K (I)	Chart	Link
DocVQA	50K	12,767 (I)	Text, Table, Chart	Link
VisualMRC	30,562	10,197 (I)	Text, Table, Chart	Link
TAT-DQA	16,558	2,758 (D)	Text, Table, Chart	Link
InfoVQA	30K	5.4K (I)	Text, Table, Chart	Link
ChartQA	23.1K	17.1K (I)	Chart	Link
ScienceQA	21K	7,803 (I)	Text, Table, Chart	Link
DUDE	41,491	4,974 (D)	Text, Table, Chart	Link
SlideVQA	52K	14.5K (I)	Slide	Link
ArXivQA	100K	16.6K (D)	Text, Table, Chart	Link
MMLongBench-Doc	1,062	130 (D)	Text, Table, Chart, Slide	Link
PaperTab	393	307 (D)	Text, Table	Link
FetaTab	1,023	878 (D)	Table	Link
SPIQA	27K	25.5K (D)	Table, Chart	Link
LongDocURL	2,325	396 (D)	Text, Table, Chart	Link

Multimodal RAG Benchmarks

Benchmark	#Queries	#Docs/Images	Content	Introduced By	Paper
ViDoRe	3.8K	8.3K (D)	Text, Table, Chart	ColPali	Link
VisR-Bench	471	226 (D)	Text, Table, Chart, Slide	VisR-Bench	Link
M3DocVQA	2,441	3,368 (D)	Text, Table, Chart	M3DocRAG	Link
VisDoMBench	2,271	1,277 (D)	Text, Table, Chart, Slide	VisDoMRAG	Link
ViDoSeek	1,142	300 (D)	Text, Table, Chart	ViDoRAG	Link
OpenDocVQA	206K	43K (I)	Text, Table, Chart	VDocRAG	Link
UniDoc-Bench	1.6K	70K (I)	Text, Table, Chart	UniDoc	Link
BBox-DocVQA	32K	4.4K (D)	Text, Table, Chart	BBox-DocVQA	Link

Citation

If you find this survey useful, please cite our paper:

@article{gao2025scaling,
      title={Scaling Beyond Context: A Survey of Multimodal Retrieval-Augmented Generation for Document Understanding},
      author={Sensen Gao and Shanshan Zhao and Xu Jiang and Lunhao Duan and Yong Xien Chng and Qing-Guo Chen and Weihua Luo and Kaifu Zhang and Jia-Wang Bian and Mingming Gong},
      year={2025},
      eprint={2510.15253},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2510.15253},
}

Contact

For questions or suggestions, feel free to open an issue or contact Sensen Gao.