Catalog
SYS-4SYSP1/MSpec-levelPROPOSED

引入 A-Mem 的 Zettelkasten 卡片演化 Schema:让 Derived Layer 有坚实的图谱结构

Evaluation modality

Spec-level

A spec-motivation / governance borrow. Evaluated by spec review + contract tests, not A/B or ablation.

Primary owner
Phase-A verdict
Shadow profile
Source papers
A-Mem 2025
Specs
docs/specs/continuum-memory.md

Blind spot (现状盲点)

CMA 组关注了记忆的探针测试和漂移监控,DM-6 关注了记忆的价值评估,但缺乏**记忆到底长什么样**的具体数据结构。Derived Layer(派生索引层)如果只有向量数据库,将难以支撑复杂的图谱推理。

Adoptable suggestions (可落地动作)

  1. 1.在 [`docs/specs/continuum-memory.md`](../specs/continuum-memory.md) 中,借鉴 A-Mem 的 Attribute Schema。PROPOSED

    Not a runnable A/B candidate — evaluated by the path above, not ablation.

  2. 2.为 Episodic 到 Semantic 的演化路径建立 Zettelkasten 式的动态卡片系统:新记忆自动生成属性、关联,并演化已有记忆。PROPOSED

    Not a runnable A/B candidate — evaluated by the path above, not ablation.

  3. 3.**注意约束**:仅借鉴其数据结构 Schema 和演化逻辑,**绝不引入 LLM-as-curator** 进行无约束的图谱突变(遵守 CMA-1 的只读/受控原则)。PROPOSED

    Not a runnable A/B candidate — evaluated by the path above, not ablation.

Traceability

No plugins / runs linked yet. Scaffold a suggestion to start.

Expected benefit (预期收益)

- 让 Derived Layer 有更坚实的图谱结构和明确的字段 Schema,而不仅仅是扁平的文本块。 - 为多跳联想(CMA-1)和记忆溯源提供结构化数据支撑。

Cited paper (引用论文)

**A3-05. A-Mem: Agentic Memory. arXiv:2502.12110, 2025.** - 关键观点:Zettelkasten 式动态卡片系统,新记忆自动生成属性、关联、演化已有记忆,是 episodic→semantic 演化路径的绝佳蓝本。 ---