Retrieval-augmented generation (RAG) has demonstrated strong performance in single-hop question answering (QA) by integrating external knowledge into large language models (LLMs). However, its effectiveness remains limited in multi-hop QA, which demands both stable reasoning and factual consistency. Existing approaches often provide partial solutions, addressing either reasoning trajectory stability or factual verification, but rarely achieving both simultaneously. To bridge this gap, we propose PAR-RAG, a three-stage Plan-then-Act-and-Review framework inspired by the PDCA cycle. PAR-RAG incorporates semantic complexity as a unifying principle through three key components: (i) complexity-aware exemplar selection guides plan generation by aligning decomposition granularity with question difficulty, thereby stabilizing reasoning trajectories; (ii) execution follows a structured retrieve-then-read process; and (iii) dual verification identifies and corrects intermediate errors while dynamically adjusting verification strength based on question complexity: emphasizing accuracy for simple queries and multi-evidence consistency for complex ones. This cognitively inspired framework integrates theoretical grounding with practical robustness. Experiments across diverse benchmarks demonstrate that PAR-RAG consistently outperforms competitive baselines, while ablation studies confirm the complementary roles of complexity-aware planning and dual verification. Collectively, these results establish PAR-RAG as a robust and generalizable framework for reliable multi-hop reasoning.

翻译：检索增强生成（RAG）通过将外部知识整合到大型语言模型（LLMs）中，已在单跳问答（QA）任务中展现出强大性能。然而，其在需要稳定推理与事实一致性的多跳问答中，有效性仍然有限。现有方法通常提供部分解决方案，或关注推理轨迹的稳定性，或关注事实核查，但很少能同时实现两者。为弥合这一差距，我们提出PAR-RAG，一个受PDCA循环启发的三阶段“规划-执行-审查”框架。PAR-RAG将语义复杂性作为统一原则，通过三个关键组件实现：（i）复杂度感知的示例选择通过使分解粒度与问题难度对齐来指导规划生成，从而稳定推理轨迹；（ii）执行遵循结构化的“检索-阅读”流程；（iii）双重验证识别并纠正中间错误，同时根据问题复杂度动态调整验证强度：对简单查询强调准确性，对复杂查询则强调多证据一致性。这一受认知启发的框架将理论依据与实践鲁棒性相结合。在多个基准测试上的实验表明，PAR-RAG始终优于竞争基线方法，而消融研究则证实了复杂度感知规划与双重验证的互补作用。总体而言，这些结果确立了PAR-RAG作为一个鲁棒且可泛化的框架，用于实现可靠的多跳推理。