{"ID":2859219,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.05691","arxiv_id":"2510.05691","title":"DecEx-RAG: Boosting Agentic Retrieval-Augmented Generation with Decision and Execution Optimization via Process Supervision","abstract":"Agentic Retrieval-Augmented Generation (Agentic RAG) enhances the processing capability for complex tasks through dynamic retrieval and adaptive workflows. Recent advances (e.g., Search-R1) have shown that outcome-supervised reinforcement learning demonstrate strong performance. However, this approach still suffers from inefficient exploration, sparse reward signals, and ambiguous global reward feedback. To address these challenges, we propose DecEx-RAG, which models RAG as a Markov Decision Process (MDP) incorporating decision-making and execution, while introducing an efficient pruning strategy to optimize data expansion. Through comprehensive process-level policy optimization, DecEx-RAG significantly enhances the autonomous task decomposition, dynamic retrieval, and high-quality answer generation capabilities of large language models (LLMs). Experiments show that DecEx-RAG achieves an average absolute performance improvement of $6.2\\%$ across six datasets, significantly outperforming existing baselines. Moreover, the pruning strategy improves data construction efficiency by nearly $6 \\times$, providing an efficient solution for process-supervised RAG training. The code is available at https://github.com/sdsxdxl/DecEx-RAG.","short_abstract":"Agentic Retrieval-Augmented Generation (Agentic RAG) enhances the processing capability for complex tasks through dynamic retrieval and adaptive workflows. Recent advances (e.g., Search-R1) have shown that outcome-supervised reinforcement learning demonstrate strong performance. However, this approach still suffers fro...","url_abs":"https://arxiv.org/abs/2510.05691","url_pdf":"https://arxiv.org/pdf/2510.05691v1","authors":"[\"Yongqi Leng\",\"Yikun Lei\",\"Xikai Liu\",\"Meizhi Zhong\",\"Bojian Xiong\",\"Yurong Zhang\",\"Yan Gao\",\"Yi Wu\",\"Yao Hu\",\"Deyi Xiong\"]","published":"2025-10-07T08:49:22Z","proceeding":"cs.CL","tasks":"[\"cs.CL\"]","methods":"[\"RAG\",\"Reinforcement Learning\",\"Large Language Model\",\"Language Model\",\"LoRA\"]","has_code":false,"code_links":[{"ID":608619,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2859219,"paper_url":"https://arxiv.org/abs/2510.05691","paper_title":"DecEx-RAG: Boosting Agentic Retrieval-Augmented Generation with Decision and Execution Optimization via Process Supervision","repo_url":"https://github.com/sdsxdxl/DecEx-RAG","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}