[论文解读] Scene-Aware Memory Discrimination: Deciding Which Personal Knowledge Stays

Yijie Zhong, Mengying Guo+4 more|arXiv (Cornell University)|Feb 12, 2026

Personal Information Management and User Behavior被引用 0

一句话总结

SAMD 使用门控单元与聚类提示模块进行场景感知的记忆区分，在提升记忆质量的同时降低计算成本，并且在直接与间接评估中表现出色。

ABSTRACT

Intelligent devices have become deeply integrated into everyday life, generating vast amounts of user interactions that form valuable personal knowledge. Efficient organization of this knowledge in user memory is essential for enabling personalized applications. However, current research on memory writing, management, and reading using large language models (LLMs) faces challenges in filtering irrelevant information and in dealing with rising computational costs. Inspired by the concept of selective attention in the human brain, we introduce a memory discrimination task. To address large-scale interactions and diverse memory standards in this task, we propose a Scene-Aware Memory Discrimination method (SAMD), which comprises two key components: the Gating Unit Module (GUM) and the Cluster Prompting Module (CPM). GUM enhances processing efficiency by filtering out non-memorable interactions and focusing on the salient content most relevant to application demands. CPM establishes adaptive memory standards, guiding LLMs to discern what information should be remembered or discarded. It also analyzes the relationship between user intents and memory contexts to build effective clustering prompts. Comprehensive direct and indirect evaluations demonstrate the effectiveness and generalization of our approach. We independently assess the performance of memory discrimination, showing that SAMD successfully recalls the majority of memorable data and remains robust in dynamic scenarios. Furthermore, when integrated into personalized applications, SAMD significantly enhances both the efficiency and quality of memory construction, leading to better organization of personal knowledge.

研究动机与目标

• 推动记忆区分以为个性化应用构建简洁、高质量的个人记忆。
• 提出一个在冻结LLM下运行的场景感知记忆区分框架（SAMD）。
• 设计GUM以使用记忆场景显著性筛选非难忘数据，使用CPM通过意图-场景聚类来定义记忆标准。
• 通过在多样数据集上的直接与间接评估，展示SAMD的有效性与鲁棒性。

提出的方法

• 将记忆区分引入二元决策：在上下文中是否应将每句话存储为记忆。
• Gating Unit Module (GUM) 使用基于记忆场景的标识符（由显著词汇与多视角角色扮演构成）来筛选非难忘数据。
• Cluster Prompting Module (CPM) 通过将用户意图与意图-场景亲和矩阵聚类来定义应记住的内容，并为每个簇构建区分规则。
• 使用冻结的LLM应用区分规则；以最小成本更新对新记忆场景与意图的可访问性。
• 直接评估记忆区分的准确性、召回率与效率；间接评估对搭载记忆能力的代理在记忆构建中的影响。

实验结果

研究问题

• RQ1 SAMD 在大规模下能否准确区分可记忆与不可记忆的数据？
• RQ2 场景感知的区分是否在不同记忆场景与意图下提升记忆质量与效率？
• RQ3 SAMD 在不对LLM进行微调的情况下对新记忆场景或意图有多大适应性？
• RQ4 SAMD 对下游记忆辅助应用在速度与质量方面的影响如何？

主要发现

• 带有意图聚类的SAMD在关键数据集上比基线方法实现约3倍的更快区分和超过20%的准确率提升。
• SAMD能保留超过90%的可记忆数据，整体性能提升超过20%。
• GUM在早期有效筛选非难忘数据，降低不必要的LLM使用和计算成本。
• CPM的意图-场景聚类使区分规则覆盖多样表达，同时尽量缩小跨场景的记忆缺口。
• 当用户意图不完整、错误或不可用时，SAMD仍然鲁棒，并能适应不断变化的应用需求。