MC-Search：基于结构化长推理链评估和增强多模态代理搜索
MC-Search: Evaluating and Enhancing Multimodal Agentic Search with Structured Long Reasoning Chains
AI总结：
MC-Search是首个针对代理MM-RAG的基准，通过结构化长推理链评估和增强多模态代理搜索，揭示了系统性问题并改进了模型的规划和检索能力。
摘要：
随着对分步、跨模态和知识引导推理需求的增加，多模态大语言模型（MLLMs）正在从传统的固定检索-生成范式向更复杂的代理多模态检索增强生成（MM-RAG）发展。然而，现有基准主要集中在简化问答和短检索链上，使得适应性规划和多模态推理研究不足。我们提出了MC-Search，首个针对代理MM-RAG的基准，包含五个代表性推理结构的长、分步注释推理链。每个示例指定子问题、检索模态、支持事实和中间答案，通过HAVE（分步证据归属和验证）确保准确性，产生3333个高质量示例，平均3.7步。除了回答准确性，MC-Search引入了新的过程级指标用于推理质量、分步检索和规划准确性。通过开发统一的代理MM-RAG流程，我们评估了六个领先的MLLMs，揭示了系统性问题，如过度和不足检索以及模态不一致的规划。最后，我们引入Search-Align，一个利用验证推理链的过程监督微调框架，证明我们的数据不仅能够实现忠实评估，还能提高开源MLLMs的规划和检索准确性。
英文摘要：
With the increasing demand for step-wise, cross-modal, and knowledge-grounded reasoning, multimodal large language models (MLLMs) are evolving beyond the traditional fixed retrieve-then-generate paradigm toward more sophisticated agentic multimodal retrieval-augmented generation (MM-RAG). Existing benchmarks, however, mainly focus on simplified QA with short retrieval chains, leaving adaptive planning and multimodal reasoning underexplored. We present MC-Search, the first benchmark for agentic MM-RAG with long, step-wise annotated reasoning chains spanning five representative reasoning structures. Each example specifies sub-questions, retrieval modalities, supporting facts, and intermediate answers, with fidelity ensured by HAVE (Hop-wise Attribution and Verification of Evidence), resulting in 3,333 high-quality examples averaging 3.7 hops. Beyond answer accuracy, MC-Search introduces new process-level metrics for reasoning quality, stepwise retrieval and planning accuracy. By developing a unified agentic MM-RAG pipeline, we benchmark six leading MLLMs and reveal systematic issues such as over- and under-retrieval and modality-misaligned planning. Finally, we introduce Search-Align, a process-supervised fine-tuning framework leveraging verified reasoning chains, showing that our data not only enables faithful evaluation but also improves planning and retrieval fidelity in open-source MLLMs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
