发表机构
Emory University（埃默里大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对大规模异构企业知识库检索难题，提出含动态多级层级等机制的MEMONDEMAND系统，在EnterpriseRAG-Bench等数据集上优于现有最优方法，是高效可扩展的企业内存解决方案。
AI 中文摘要
企业知识库规模庞大、异构且持续更新，当需同时支持高效访问、来源可靠的证据及跨查询适配时，检索工作会变得困难。企业内存将检索范围扩展至模型上下文之外，但现有系统未在该规模下协同解决特定集合的层级构建、低成本路由、详细证据加载及工作负载感知的内存更新问题。我们提出MEMONDEMAND（On-Demand Memory的缩写），这是一款内存管理系统，包含三个协同机制：为每个集合确定抽象结构和深度的动态多级层级；在每个层级分离精简路由与详细证据的双内存；在有限活跃状态预算下更新节点优先级的按需内存提升。在EnterpriseRAG-Bench上，MEMONDEMAND在从10M token到完整618M token集合的所有评估规模下，均优于已发表的最优LB#1结果，在10M时提升12.23%，在618M时提升4.66%。在FinanceBench、HotpotQA和FRAMES上的结果进一步显示其在金融、多跳及事实检索场景下的优异性能。综合这些结果，MEMONDEMAND被确立为适用于不同数据规模、领域及证据需求的超大型企业知识库的准确、高效且可扩展的内存解决方案。我们的代码可在https URL xfab-xinyuansong/MemOnDemand.git获取。
英文摘要
Enterprise repositories are large, heteroge- neous, and continuously updated, making re- trieval difficult when efficient access, source- faithful evidence, and cross-query adaptation must be supported together. Enterprise mem- ory extends retrieval beyond the model con- text, but existing systems do not jointly address collection-specific hierarchy construction, low- cost routing, detailed evidence loading, and workload-aware memory updates at this scale. We introduce MEMONDEMAND, short for On- Demand Memory, a memory management sys- tem with three coordinated mechanisms: a dy- namic multi-level hierarchy that determines the abstraction structure and depth for each col- lection, dual memory at every hierarchy level that separates distilled routing from detailed evidence, and on-demand memory promotion that updates node priority under a bounded active-state budget. On EnterpriseRAG-Bench, MEMONDEMAND outperforms the strongest published LB#1 result at every evaluated scale from 10M tokens through the complete 618M- token collection, with gains of 12.23% at 10M and 4.66% at 618M. Results on FinanceBench, HotpotQA, and FRAMES further show strong performance across financial, multi-hop, and fact-retrieval settings. Together, these results establish MEMONDEMAND as an accurate, ef- ficient, and scalable memory solution for very large enterprise repositories across data scales, domains, and evidence requirements. Our code is available at https://github.com/ xfab-xinyuansong/MemOnDemand.git.
