发表机构
C-FAIR & School of Software, Shandong University; SKLCCSE Lab, Beihang University; Department of Data Science, City University of Hong Kong（C-FAIR与山东大学软件学院; 北京航空航天大学SKLCCSE实验室; 香港城市大学数据科学系）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对基于LLM的智能体内存管理问题，提出价值感知内存管理系统MemLens，通过端到端交互式分析仪表板展示内存生命周期，经学习助手应用程序帮助用户比较策略，可实现高效、可解释和个性化的长期内存管理。
AI 中文摘要
最近，内存管理已成为基于大语言模型（LLM）的智能体的关键基础设施，因为它直接影响长期推理、个性化响应和知识重用。然而，现有的LLM内存系统通常采用粗粒度（与效用无关）的方式，统一处理异构的用户-LLM交互记录，导致冗余和低影响的记录持续存在于内存存储库中。为应对这一挑战，我们提出了MemLens，一种将内存记录视为一等数据对象的价值感知内存管理系统。MemLens提供了一个端到端的交互式分析仪表板，展示完整的内存生命周期，包括Shapley风格的内存评估、价值感知存储和内存辅助响应。通过一个学习助手应用程序，该系统使用户能够检查内存值、可视化分层内存结构，并在响应质量、检索延迟和令牌消耗方面比较各种内存管理策略。因此，我们的MemLens可以作为一个高效、可解释和个性化的基于LLM的智能体的长期内存管理系统。
英文摘要
Recently, memory management has become a key infrastructure for LLM-based agents, as it directly affects long-horizon reasoning, personalized responses, and knowledge reuse. However, existing LLM memory systems typically adopt a coarse-grained (utility-agnostic) manner that treats heterogeneous user-LLM interaction records uniformly, leading to redundant and low-impact records persisting in the memory repository. To address this challenge, we present MemLens, a value-aware memory management system that takes memory records as first-class data objects. MemLens provides an end-to-end interactive analytics dashboard that exposes the complete memory lifecycle, including Shapley-style memory evaluation, value-aware storage, and memory-assisted response. Through a study-copilot application, the system enables users to inspect memory values, visualize hierarchical memory structures, and compare various memory management strategies in terms of response quality, retrieval latency, and token consumption. Therefore, our MemLens can serve as an efficient, interpretable, and personalized long-term memory management system for LLM-based agents.
