DMAP: Human-Aligned Structural Document Map for Multimodal Document Understanding

2026-01-26

AI Summary

This work addresses the limitation of existing multimodal document question answering methods, which often neglect the hierarchical structure and logical-spatial relationships among document elements, leading to incoherent understanding. To overcome this, the authors propose a Structured Document Map (DMAP)—a novel document-level structural representation aligned with human cognitive processes. DMAP organizes heterogeneous elements such as text and figures into a hierarchical graph via a structure-semantic understanding agent, followed by iterative, structure-aware, evidence-driven reasoning performed by a reflective reasoning agent. Integrated with multimodal retrieval-augmented generation (RAG), the proposed framework significantly improves retrieval accuracy, reasoning consistency, and multimodal comprehension on the MMDocQA benchmark, outperforming conventional RAG approaches.

Technology Category

Machine Learning: Multimodal LearningComputer Vision: Multi-modal VisionNatural Language Processing: Question Answering

Application Category

Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAGGraph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphsSemantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Existing multimodal document question-answering (QA) systems predominantly rely on flat semantic retrieval, representing documents as a set of disconnected text chunks and largely neglecting their intrinsic hierarchical and relational structures. Such flattening disrupts logical and spatial dependencies - such as section organization, figure-text correspondence, and cross-reference relations, that humans naturally exploit for comprehension. To address this limitation, we introduce a document-level structural Document MAP (DMAP), which explicitly encodes both hierarchical organization and inter-element relationships within multimodal documents. Specifically, we design a Structured-Semantic Understanding Agent to construct DMAP by organizing textual content together with figures, tables, charts, etc. into a human-aligned hierarchical schema that captures both semantic and layout dependencies. Building upon this representation, a Reflective Reasoning Agent performs structure-aware and evidence-driven reasoning, dynamically assessing the sufficiency of retrieved context and iteratively refining answers through targeted interactions with DMAP. Extensive experiments on MMDocQA benchmarks demonstrate that DMAP yields document-specific structural representations aligned with human interpretive patterns, substantially enhancing retrieval precision, reasoning consistency, and multimodal comprehension over conventional RAG-based approaches. Code is available at https://github.com/Forlorin/DMAP

Problem

• multimodal document understanding
• document structure
• hierarchical organization
• relational dependencies
• document question answering

Innovation

• structural document representation
• hierarchical reasoning
• multimodal document understanding
• reflective reasoning agent
• human-aligned schema