MAGE-RAG: Multigranular Adaptive Graph Evidence for Agentic Multimodal RAG in Long-Document QA

2026-06-14

AI Summary

This work addresses the challenge in existing retrieval-augmented generation (RAG) approaches of effectively integrating visual layout cues with the efficient localization of sparse, scattered evidence in long-document multimodal question answering. The authors propose a multi-granularity adaptive graph-based evidence framework that constructs, offline, a multi-relational evidence graph incorporating page- and element-level nodes along with their hierarchical, spatial, and semantic relationships. During inference, an evidence controller dynamically activates, searches, and prunes this graph under computational budget constraints to produce a compact multimodal subgraph tailored for large vision-language models. This approach achieves the first query-driven, dynamic construction of multi-granularity evidence subgraphs, flexibly balancing evidence coverage, noise suppression, and reasoning cost. It attains state-of-the-art accuracy of 52.75 on LongDocURL and 53.26 on MMLongBench-Doc (F1 = 51.19), substantially outperforming current text- and page-level visual RAG baselines.

Abstract

Long-document multimodal question answering requires a system to locate sparse evidence in long PDFs and integrate clues from text, tables, images, charts, and complex layouts. Existing RAG methods mostly rely on fixed Top-k retrieval over text chunks or pages. Text retrieval can compress the context but often loses visual and layout information; page-level visual retrieval preserves the original page, yet it also sends large irrelevant regions to the reader, leading to a static trade-off among evidence coverage, noise, and inference cost. This paper proposes MAGE-RAG, a multigranular adaptive graph evidence framework for long-document multimodal QA. MAGE-RAG uses page retrieval as the entry point for query-time evidence construction. Offline, it builds an evidence graph with page nodes and element nodes, encoding containment, reading order, layout adjacency, section hierarchy, and semantic-neighbor relations. At query time, an online evidence controller iteratively activates, opens, searches, and prunes evidence under explicit budgets. The resulting evidence subgraph is then rendered into structured multimodal reader input, allowing the LVLM to consume compact and relevant evidence within a limited context. On LongDocURL and MMLongBench-Doc, we establish a unified comparison and analysis protocol covering Direct MLLM, Text RAG, Page-level Visual RAG, and Graph/Agentic RAG. Experiments show that MAGE-RAG achieves 52.75 overall accuracy on LongDocURL, and 53.26 accuracy with 51.19 F1 on MMLongBench-Doc. Fine-grained breakdowns, budget-performance curves, ablations, and trace-based analysis further show that query-time evidence subgraph construction can balance dispersed evidence coverage with context-noise control. Our code is available at https://github.com/laonuo2004/MAGE-RAG.git.

Problem

• long-document multimodal QA
• evidence retrieval
• multimodal RAG
• visual and layout information
• context-noise trade-off

Innovation

• Multigranular Retrieval
• Adaptive Graph Evidence
• Agentic RAG
• Multimodal QA
• Long-document Understanding