From Naive RAG to Deep Agentic Retrieval: An Evolving Context Engineering Pipeline for Regulatory Compliance
从朴素检索增强生成到深度智能检索:用于合规监管的不断演进的上下文工程管道
专题命中 检索器与排序 :RAG(title,summary_cn);retrieval-augmented generation(abstract);hybrid retrieval(abstract);分类 cs.IR、cs.CL、cs.AI
AI总结 研究针对RAG朴素实现的局限,追溯安大略发电公司检索管道演变,历经多阶段形成PEA-CAE架构,表明上下文工程对监管语料库更具优势,深度智能检索进展反映经典思想,迭代证据获取等渐成企业问答基础。
Comments Accepted at the 2026 IEEE the 14th International Conference on Smart Energy Grid Engineering (SEGE 2026)