Graph Retrieval-Augmented Generation (Graph-RAG) enhances multihop question answering by organizing corpora into knowledge graphs and routing evidence through relational structure. However, practical deployments face two persistent bottlenecks: (i) mixed-difficulty workloads where one-size-fits-all retrieval either wastes cost on easy queries or fails on hard multihop cases, and (ii) extraction loss, where graph abstraction omits fine-grained qualifiers that remain only in source text. We present A2RAG, an adaptive-and-agentic GraphRAG framework for cost-aware and reliable reasoning. A2RAG couples an adaptive controller that verifies evidence sufficiency and triggers targeted refinement only when necessary, with an agentic retriever that progressively escalates retrieval effort and maps graph signals back to provenance text to remain robust under extraction loss and incomplete graphs. Experiments on HotpotQA and 2WikiMultiHopQA demonstrate that A2RAG achieves +9.9/+11.8 absolute gains in Recall@2, while cutting token consumption and end-to-end latency by about 50% relative to iterative multihop baselines.

翻译：图检索增强生成（Graph-RAG）通过将语料库组织为知识图谱并利用关系结构进行证据路由，提升了多跳问答的性能。然而，实际部署面临两个持续存在的瓶颈：（i）混合难度的工作负载，其中“一刀切”的检索方式要么在简单查询上浪费成本，要么在困难的多跳案例上失败；（ii）提取损失，即图抽象过程省略了仅保留在源文本中的细粒度限定信息。本文提出A2RAG，一种面向成本感知与可靠推理的自适应智能体化GraphRAG框架。A2RAG结合了一个自适应控制器——用于验证证据充分性并仅在必要时触发针对性细化，以及一个智能体检索器——逐步提升检索力度并将图谱信号映射回溯源文本，从而在提取损失与图谱不完整的情况下保持鲁棒性。在HotpotQA和2WikiMultiHopQA上的实验表明，相较于迭代式多跳基线方法，A2RAG在Recall@2指标上实现了+9.9/+11.8的绝对提升，同时将令牌消耗和端到端延迟降低了约50%。