Skip to content

Knowledge Classification Standard

Purpose

This document defines how organizational knowledge should be categorized and systematized for both human access and AI-first knowledge systems. Understanding these classifications is foundational for building an effective knowledge architecture.

1. Types of Organizational Knowledge

Explicit Knowledge

Easily codified and stored knowledge.

Examples System Goal
Standard Operating Procedures (SOPs) Accessibility
Official Product Documentation Consistency
Company Policy Manuals
Wiki/Knowledge Base content

Treatment: Directly powers the Knowledge Graph via structured frontmatter and metadata.

Implicit Knowledge

Knowledge derived from explicit knowledge but requiring context to use.

Examples System Goal
Database schemas Connective Tissue
APIs (via Knowledge Graph)
Code repositories

Treatment: Code embedding and integration with the Knowledge Graph to map entities (e.g., "Customer ID field") to documentation.

Tacit/Dark Knowledge

Uncodified, often informal knowledge derived from experience, conversations, and context.

Examples System Goal
Vendor/Partner Q&A Discovery
Team Meeting Notes Extraction
Internal Email Threads (for the AI layer)
Slack/Chat discussions
Project Decisions

Treatment: Powers the RAG (Retrieval-Augmented Generation) layer and is linked to the Knowledge Graph via entity extraction.

2. Where Different Knowledge Types Sit

Knowledge Type Current Location (Human-Centric) AI Layer Treatment
Explicit (Official Docs) Wiki, Documentation System Indexed directly; integrated into the Knowledge Graph via frontmatter/metadata
Tacit (Vendor Q&A, Meetings) Google Docs, Email, Drive, Confluence, Transcripts RAG. AI searches raw content for context, then uses KG for grounding
Implicit (API/Code) Code Repositories, Databases Code Embedding and integration with KG to map entities to documentation

3. Systematizing Tacit Knowledge for AI

The true value of an AI-first knowledge system comes from moving beyond static documentation to capture the "why" and "how" of daily operations.

The Role of Access & RAG

The challenge isn't just access; it's contextual access.

Access Control:

  • Granular access respects privacy (e.g., only finance team sees confidential vendor negotiation notes)

Retrieval-Augmented Generation (RAG):

  • You don't need to manually re-write every meeting note into a structured format
  • Index raw Google Docs and transcripts using Vector Embeddings
  • When a user asks "Why did we choose Vendor X over Vendor Y?", the AI:
    1. Searches the Vector Database (containing embedded meeting notes)
    2. Pulls relevant paragraphs
    3. Uses those snippets as context to generate a precise, grounded answer

Connection to the Knowledge Graph

While raw documents don't sit in the Knowledge Graph (KG), they should be linked to the KG.

Entity Linking:

  • AI scans a meeting transcript and automatically identifies key entities (e.g., Project Alpha, Vendor Acme, Employee Jane Doe)

Metadata Bridge:

  • These entities become nodes in your KG
  • The original Google Doc/Meeting Transcript becomes a relationship property on those nodes
  • Example: "Project Alpha discussed_in Document_ID_123"
  • This allows the AI to traverse the KG and find the context in the unstructured data

4. Summary: Integrated Knowledge Architecture

Layer Knowledge Type Purpose
Knowledge Graph Explicit Answer "What is the policy?"
RAG Layer Tacit Answer "Why did we make that decision?"
Code Embeddings Implicit Connect systems to documentation

Integration Flow:

┌─────────────────────────────────────────────────────────────┐
│                    User Question                             │
└─────────────────────────────────────────────────────────────┘
                            │
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                     AI Query Layer                          │
├─────────────────┬─────────────────┬─────────────────────────┤
│ Knowledge Graph │   RAG Layer     │    Code Embeddings      │
│ (Explicit)      │   (Tacit)       │    (Implicit)           │
├─────────────────┼─────────────────┼─────────────────────────┤
│ Wiki/Docs       │ Meeting Notes   │ Database Schemas        │
│ SOPs            │ Email Threads   │ API Definitions         │
│ Policies        │ Vendor Q&A      │ Code Repositories       │
└─────────────────┴─────────────────┴─────────────────────────┘

5. Metaknowledge: Documenting the System Itself

This document is an example of metaknowledge—documentation that describes how the knowledge system itself operates.

Placement Strategy

When creating system architecture documentation:

  1. Explicit Knowledge Base: Store as structured Markdown with proper frontmatter
  2. Knowledge Graph: Extract key concepts as nodes with relationships:
    • Explicit Knowledge → IS_STORED_IN → Wiki
    • Tacit Knowledge → IS_PROCESSED_BY → RAG Layer
    • RAG Layer → USES → Vector Database
  3. Vector Embedding: Index for RAG to answer philosophical questions about the system

Metadata Template for Metaknowledge

---
knowledge_type: Metaknowledge
domain: System Architecture
audience: Internal/Developer/PM
key_entities:
  - Explicit Knowledge
  - Tacit Knowledge
  - RAG
  - Knowledge Graph
---