TrawMem: 重构智能体记忆为任务导向的角色自适应工作区
TrawMem: Reframing Agent Memory as Task-Oriented Role-Adaptive Workspaces
Mingyu Zhang · Zixu Li · Zhiheng Fu · Zhiwei Chen · Jinhe Bi · Yupeng Hu
中文摘要
大语言模型(LLM)智能体依赖记忆系统利用长期交互，但现有系统通常基于查询相关性检索记忆并作为上下文证据。然而，相关性无法表征记忆对任务解决的贡献：同一记忆在不同任务中可能扮演不同角色，如直接支持答案或链接语义相关性较弱但信息必要的互补记忆。因此，将智能体记忆系统重构为任务导向的角色自适应工作区，按任务导向的角色和关系选择并组织记忆。基于此视角，提出 TrawMem，将交互历史组织为局部线程，通过分配角色并联合选择互补记忆构建任务导向工作区。在 LoCoMo 和 MemGallery 上使用四个开源和闭源骨干模型的实验显示，相比代表性记忆系统有一致提升，宏平均 F1(macro-F1) 提升 2.23-3.14 分。在 LoCoMo 上，TrawMem 相比 SimpleMem 在 Qwen3.5-9B 和 DeepSeek-V4-Flash 上分别提升 LLM-Judge 分数 4.07 和 4.31 个百分点，同时分别减少约 12.8% 和 5.1% 的上下文 token 需求。
关键要点
- 01现有记忆系统仅基于查询相关性检索，无法区分记忆在任务中的具体角色与贡献方式。
- 02提出 TrawMem，将记忆组织为局部线程，通过分配角色并联合选择互补记忆构建任务导向工作区。
- 03在 LoCoMo 和 MemGallery 基准上，TrawMem 相比代表性系统宏平均 F1 提升 2.23-3.14 分。
- 04在 LoCoMo 上，TrawMem 相比 SimpleMem 提升 LLM-Judge 分数 4.07-4.31 个百分点，且减少上下文 token 需求。
解读
尚无解读。
原始英文摘要
arXiv:2610.36918v1 Announce Type: new Abstract: Large language model (LLM) agents rely on memory systems to leverage long-term interactions, yet existing systems typically retrieve memories based on their relevance to the current query and use them as contextual evidence. However, relevance alone does not characterize how a memory contributes to task solving: the same memory may serve different roles across tasks, such as directly supporting the answer or linking to complementary memories with weaker semantic relevance but necessary information. We therefore reframe agent memory systems as Task-Oriented Role-Adaptive Workspaces, where memories are selected and organized by their task-oriented roles and relations. Based on this view, we propose TrawMem, which organizes interaction histories into local threads and constructs task-oriented workspaces by assigning roles and jointly selecting complementary memories. Experiments on LoCoMo and MemGallery with four open- and closed-source backbones show consistent improvements over representative memory systems, with macro-F1 gains of 2.23-3.14 points. On LoCoMo, TrawMem improves the LLM-Judge score over SimpleMem by 4.07 and 4.31 percentage points on Qwen3.5-9B and DeepSeek-V4-Flash, respectively, while requiring approximately 12.8% and 5.1% fewer context tokens, respectively.