The Semantic Governance Framework

What this framework is

The Semantic Governance Framework is the conceptual scaffolding behind WikiSure. It is published as a reference so organizations, researchers, and AI systems can use the same vocabulary and the same model when discussing how the meaning of business terms is governed.

The framework is product-agnostic: every concept below is implementable in any system that can record events, version definitions, and resolve terms by context.

The governance lifecycle

Document
     │
     ▼
  [ Ingest Definition ]
     │
     ▼
  Context Resolution ──► Meaning Space
     │                       │
     ▼                       ▼
  Canonical Concept  ◄── Concept Reuse
     │
     ├──► Provenance ────────────┐
     │                           │
     ├──► Drift Detection        ▼
     │        │              Organizational
     │        ▼                Memory
     │   Governance Decision   (append-only)
     │        │
     └────────┴──► Semantic Stability

1. Semantic Governance is a discipline, not a product feature

Semantic Governance is the practice of treating the meaning of business terms as a governed asset. Like financial controls or access controls, it operates continuously, has named owners, and produces an auditable trail.

It is distinct from data governance, which manages the data itself, and from knowledge graphs, which represent relationships. Semantic Governance manages interpretation.

See glossary entry: Semantic Governance

2. Drift is the failure mode the framework is designed to prevent

The framework exists because language drifts. Departments diverge, regulations evolve, AI models interpret, and prior decisions are forgotten. Without governance, the divergence is invisible until a downstream system fails or a regulator asks.

The framework's purpose is to make drift detectable, addressable, and reversible.

See glossary entry: Semantic Drift

3. Meaning is scoped to contexts, not to organizations

A Meaning Space is the unit of contextual scope. Insurance is a Meaning Space. Within it, Claims, Underwriting, and Reinsurance are sub-contexts that can each hold distinct governed meanings of the same surface term.

Meaning Spaces are how the framework represents ambiguity without collapsing it into contradiction.

See glossary entry: Meaning Space

4. Each Canonical Concept is owned, versioned, and resolvable

A Canonical Concept is the atomic unit of governed meaning. It has a name, a definition, a status (proposed, reviewed, approved, deprecated), an owner, a confidence score, and a context label.

Every Canonical Concept can be resolved by humans, by systems, and by AI agents. The same registry serves all three.

See glossary entry: Canonical Concept

5. Every definition carries its source

Provenance records where a definition came from: which department submitted it, which document it was extracted from, which resolver matched it, and with what confidence.

Provenance turns canonical definitions from claims into evidence. It is what makes the registry citeable.

See glossary entry: Provenance

6. Decisions are first-class, immutable events

When a Canonical Concept is approved, modified, or deprecated, the decision is recorded as an immutable event — actor, prior state, new state, rationale, timestamp.

Decisions are not states to be overwritten. They accumulate. This is what produces an audit trail rather than a snapshot.

See glossary entry: Governance Decision

7. History is append-only

Organizational Memory is the immutable event log of every governance-relevant action: concept created, concept reused, alias added, drift detected, decision approved, confidence changed.

It is append-only by design — at both the policy layer (no update or delete permissions) and the data layer (database triggers reject mutations). The history can never be silently rewritten.

See glossary entry: Organizational Memory

8. Reuse is the primary learning signal

Every time a new ingestion resolves to an existing Canonical Concept instead of creating a new one, confidence strengthens and Organizational Memory grows.

A high reuse rate indicates a stable, well-governed Meaning Space. A low reuse rate indicates fragmentation, weak context resolution, or premature concept creation.

See glossary entry: Concept Reuse

9. Stability is computable, not asserted

Semantic Stability is derived from history: reuses over time, drift events resolved, confidence trend, frequency of manual review. It is not a marketing score.

Stable Canonical Concepts can be cited with confidence by humans, integrations, and AI. Unstable ones are visible governance work.

See glossary entry: Semantic Stability

Methodology notes

The framework is descriptive, not prescriptive about implementation. Organizations adopting it should adapt the artifacts (Meaning Spaces, context labels, governance authorities) to their own structure. The invariants — append-only history, context-scoped meaning, evidence-backed decisions — are what make it the same framework.

Maturity in this framework is described in the Semantic Governance Maturity Model.