Skip to content

Compiled Truth + Timeline Model

The Compiled Truth + Timeline Model is a data architecture and knowledge management strategy designed to balance the stability of current understanding with the immutability of historical records^[001-TODO__GBrain_-AI_Agent_个人知识库与混合检索引擎.md]. This model structures information into two distinct zones within a single document: a mutable "Compiled Truth" section representing the best current understanding of a topic, and an immutable "Timeline" section that serves as a chronological log of events^[001-TODO__GBrain-_AI_Agent_个人知识库与混合检索引擎.md].

This approach addresses the challenge of maintaining accurate summaries while preserving the granular history of how that understanding evolved over time^[001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md].

Structure and Format

The model defines a strict separation between the "current state" and the "history log" within a page^[001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md].

Compiled Truth (Mutable)

The top section of the document is the Compiled Truth. It represents a "living" summary that is continuously rewritten and refined as new information becomes available^[001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md]. * Purpose: To provide the current best understanding or a consolidated view of the entity (e.g., a person, company, or concept). * Behavior: It is designed to be edited and updated freely to reflect the latest knowledge.

Timeline (Immutable)

The bottom section is the Timeline. It serves as an append-only ledger of events or interactions^[001-TODO__GBrain_-AI_Agent_个人知识库与混合检索引擎.md]. * Purpose: To preserve the historical record and the provenance of facts. * Behavior: Entries are added in chronological order. Crucially, past entries are never edited or deleted, ensuring that the original context and data points remain intact even if the "Compiled Truth" above changes^[001-TODO__GBrain-_AI_Agent_个人知识库与混合检索引擎.md].

YAML Frontmatter

Documents utilizing this model typically begin with metadata headers to define the context and type of the page^[001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md]: 

---
type: person | company | concept | deal | project
title: ...
tags: [...]
---

Example Format

A typical document following this model appears as follows^[001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md]:

---
type: person
title: Jane Doe
tags: [engineer, executive]
---

# Compiled Truth
(Analysis and summary of Jane Doe's current role, impact, and relationship status.
This section is rewritten as new significant events occur.)

---

# Timeline
- 2024-01-15: Jane promoted to CTO.
- 2024-03-20: Jane met with investors at YC.
- 2024-05-10: Published a paper on AI agents.

Applications

  • AI Agent Knowledge Bases: This model is central to systems like GBrain, where AI agents need a consistent "source of truth" to answer queries (the Compiled Truth) while simultaneously logging new interactions or data points to the Timeline for future processing^[001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md].
  • Personal Knowledge Management (PKM): It helps individuals manage complex relationships or projects by maintaining a clean summary for quick reference while preserving a detailed log for auditing or memory refresh.
  • Zettelkasten: The focus on atomic notes and linking contrasts with the "Compiled Truth" summary, though both support knowledge evolution.
  • [[Event Sourcing]]: The "Timeline" section mirrors the principle of storing state as a sequence of events rather than just the current state.
  • [[Progressive Summarization]]: A technique for iteratively summarizing notes, which aligns with the refinement of the Compiled Truth section.

Sources

  • 001-TODO__GBrain_-_AI_Agent_个人知识库与混合检索引擎.md