DualG-MRAG: Decoupling Macro-Reasoning and Micro-Matching for Multimodal Retrieval-Augmented Generation
2026-07-30 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors address problems in retrieving and reasoning with information from multiple sources like images and text together. They note current methods either lose important details or bring in too much irrelevant data when trying to connect different pieces of information. Their solution, called DualG-MRAG, uses two separate graphs: one for big-picture reasoning and another for checking detailed facts. This approach helps the system find and use evidence more accurately when answering complex questions. They also introduce a new way to guide the answer generation based on the reasoning paths found in these graphs.
Multimodal Retrieval-Augmented GenerationMulti-hop reasoningGraph Neural NetworkMacro graphMicro graphEvidence recallQuery-driven message passingDynamic programming decodingTopological routingComplex question answering
Authors
Jiacheng Tao, Qingyun Sun, Haonan Yuan, Ziwei Zhang, Jianxin Li
Abstract
While Multimodal Retrieval-Augmented Generation (MM-RAG) has shown promising results, it still struggles with complex multi-hop reasoning tasks. Existing methods primarily focus on independent instance-level matching, which often fails to capture explicit relationships across modalities and documents. Although Graph-enhanced methods introduce structural modeling, they face a fundamental challenge in multimodal scenarios: incorporating fine-grained visual features leads to rapid graph expansion and retrieval noise, whereas coarse-grained representations cause the discarding of critical local evidence. To address this dilemma, we propose DualG-MRAG, a Dual-tier framework that introduces a decoupled architecture comprising Macro-reasoning and Micro-matching Graphs for Multimodal RAG. Specifically, to suppress retrieval noise by isolating global structural reasoning from fine-grained evidence matching, we construct a Macro Graph for global topological routing and a Micro Graph for precise local verification. Subsequently, to enable dynamic relevance propagation across heterogeneous evidence sources, we formulate retrieval as a query-driven message passing process via a GNN Retriever. Furthermore, to provide the generative model with coherent structural guidance, we introduce a dynamic programming decoding mechanism that extracts explicit reasoning paths directly from the GNN's forward pass, replacing the standard input of isolated document chunks. Extensive experiments demonstrate that DualG-MRAG outperforms baselines in both evidence recall and complex QA accuracy.