新闻
A corrective agentic hybrid RAG and an operations-grounded evaluation for a scientific facility
arXiv cs.AI · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- Researchers deployed APS-RAG, a retrieval-augmented generation system combining dense, sparse, and knowledge-graph search channels with a corrective agentic loop for querying scientific facility operational knowledge.
- 为何重要
- Matters for engineers at large research facilities who need to search decades of accumulated operational data including logbooks, maintenance records, and control-system information through natural language.
- 注意
- Performance gains from graph channels and corrective loops were marginal; cross-encoder reranking proved critical, and the system was evaluated on only fifty questions specific to one facility.
收听本摘要
文章节选
-->
Physics > Accelerator Physics
arXiv:2607.24663v1 (physics)
[Submitted on 27 Jul 2026]
Title: A corrective agentic hybrid RAG and an operations-grounded evaluation for a scientific facility
Authors: Rajat Sainju , Dariusz Jarosz , Hairong Shang , Michael Prince , Ryan M. Aydelott , Mathew J. Cherukara , Yine Sun , Michael D. Borland
View a PDF of the paper titled A corrective agentic hybrid RAG and an operations-grounded evaluation for a scientific facility, by Rajat Sainju and 7 other authors
View PDF HTML (experimental)
Abstract: Scientific user facilities accumulate decades of operational knowledge that no single search index covers: electronic logbooks, technical documents, internal wikis, operations chat messages, maintenance records, and live control-system data. We present APS-RAG, Advanced Photon Source Retrieval Augmented Generation, a deployed platform that makes the institutional knowledge at the Advanced Photon Source (APS) accessible to staff through natural-language queries, along with an operations-grounded evaluation. The retrieval engine fuses dense, sparse, and knowledge-graph (KG) channels with query-type-adaptive reciprocal-rank fusion, adds a corrective agentic loop, and runs a native-tool ReAct executor over a Model Context Protocol (MCP) tooling layer. We construct APS-Bench, a 50-question, question-answering (QA) dataset with auditable gold an
节选自原文。请前往来源阅读全文。
- agent
- agentic
- rag
- retrieval
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。