Canonical Ontology & Schema

Capabilities ↔ Applications ↔ Data ↔ Risks ↔ Controls ↔ Outcomes #

---

1 Purpose #

Every intelligent system needs a shared language.
EA 2.0’s canonical ontology gives that language to the enterprise — a connected schema that links business intent, technical assets, risks, and measurable outcomes in a single reasoning fabric.

This layer enables traceability from strategy → system → signal.
It turns architecture from documentation into computable knowledge.

---

2 Design Principles #

Principle	Meaning
Modular	Each domain (capabilities, apps, data, etc.) can evolve independently.
Directional	Relationships have business semantics (supports, depends_on, mitigates).
Evidence-based	Every link can carry confidence, timestamp, and source_of_truth.
Machine-reasonable	Structure optimized for Cypher / Gremlin / SPARQL queries.
Governed	Changes require stewardship approval via ServiceNow or Graph UI.

---

3 Core Node Types & Attributes #

Node	Key Attributes	Description
Capability	id, name, domain, owner, priority, maturity, kpi_ref	Represents a business function or outcome.
Application	id, name, tech_stack, lifecycle, cost, cloud_provider, criticality	Implements or supports capabilities.
Data Entity	id, schema_ref, sensitivity, retention, source_system	Logical or physical data object.
Risk	id, category, probability, impact, residual_score	Threat or weakness associated with assets.
Control	id, type, status, evidence_ref, owner, coverage	Mitigating mechanism, manual or automated.
Outcome	id, metric, target, current, variance, timeframe	Business or operational result measured.

All nodes carry:
created_at, last_seen_at, source_system, confidence_score.

---

4 Canonical Relationships #

(:Capability)-[:SUPPORTED_BY]->(:Application)
(:Application)-[:USES]->(:DataEntity)
(:DataEntity)-[:HAS_RISK]->(:Risk)
(:Risk)-[:MITIGATED_BY]->(:Control)
(:Control)-[:INFLUENCES]->(:Outcome)
(:Outcome)-[:REALIZES]->(:Capability)

These six arcs form EA 2.0’s semantic backbone.
Every connector or dashboard query ultimately traverses this path.

---

5 Extended Relationship Types #

Relationship	Use
[:OWNED_BY]	Links any node → Person/Role for accountability.
[:COSTS_TO]	Application → Outcome for value/cost mapping.
[:VIOLATES]	Risk → Control for non-compliance events.
[:FORECASTS]	Model → Outcome for predictive analytics.
[:ALIGNED_WITH]	Capability → Strategic Goal for planning views.

---

6 Example Instance (Fragment) #

(Capability: "Customer Onboarding")
   ↳ Application: "CRM Portal"
      ↳ Data Entity: "CustomerProfile"
         ↳ Risk: "PII Exposure"
            ↳ Control: "Encryption at Rest"
               ↳ Outcome: "Time-to-Onboard ≤ 5 Days"

A query like

“Show controls reducing PII risk in customer onboarding and their outcome impact”
returns this chain in seconds.

---

7 Maturity Scales (per Node Type) #

Domain	Level 0	1	2	3	4	5
Capabilities	Undefined	Named	Mapped	Measured	Optimized	AI-Simulated
Applications	Untracked	Inventoried	Rationalized	Integrated	Monitored	Autonomous
Data	Siloed	Cataloged	Tagged	Linked	Trusted	Adaptive
Risk/Control	Manual	Documented	Automated	Predictive	Dynamic	Self-Healing
Outcome	Absent	Defined	Measured	Correlated	Forecasted	Optimized

These values feed Power BI’s “Maturity Heatmap.”

---

8 Ontology Governance Model #

• Stewards own domains (Capability Steward, App Owner, Risk Manager).
• All changes logged via EA 2.0 API (PATCH / PUT with JWT).
• Graph validation ensures referential completeness.
• Weekly review board checks drift > 10 %.

Governance policies define who can create, merge, or retire nodes.

---

9 Versioning and Lineage #

Each node and relationship has a lightweight version chain:

(:Application {id:'APP123', version:3})-[:SUPERSEDES]->(:Application {version:2})

This allows time-travel queries like:

“Show architecture topology as of Q2 2024.”

---

10 Implementation Schema (Neo4j / Cosmos Gremlin) #

Key Indexes

CREATE INDEX FOR (c:Capability) ON (c.id);
CREATE INDEX FOR (a:Application) ON (a.id);
CREATE INDEX FOR (d:DataEntity) ON (d.id);
CREATE INDEX FOR (r:Risk) ON (r.id);
CREATE INDEX FOR (ctrl:Control) ON (ctrl.id);
CREATE INDEX FOR (o:Outcome) ON (o.id);

Sample Query

MATCH (c:Capability)-[:SUPPORTED_BY]->(a:Application)
WHERE c.maturity >= 3
RETURN c.name, COUNT(a) AS apps, avg(a.cost) AS avg_cost
ORDER BY apps DESC LIMIT 10;

---

11 Cross-Domain Metrics #

• Capability Coverage % = # Capabilities linked to Applications / Total Capabilities.
• Data Lineage Completeness = # DataEntities linked to Risks and Controls / Total DataEntities.
• Outcome Traceability Score = # Outcomes connected back to Capabilities / Total Outcomes.
• Governance Confidence = avg(confidence_score per relationship).

These feed the Decision Latency and Trust Index dashboards.

---

12 Benefits #

✅ Full-stack traceability from goal to control.
✅ Natural-language queries become precise because relationships carry semantics.
✅ Predictive analytics operate on clean graph paths.
✅ Lineage and audit become native features of architecture.
✅ AI assistants can reason contextually without manual mapping.

---

13 Challenges & Mitigations #

Challenge	Impact	Mitigation
Duplicate applications from multiple feeds	Fragmented graph	ID harmonization rules + hashing
Inconsistent taxonomy	Poor NLQ accuracy	Central term store governance
High relationship volume	Query latency	Graph partitioning and index tuning
Rapid schema evolution	Model breakage	Schema registry with version migration scripts

---

14 Visualization Views #

• Capability Dependency Map – force layout showing business flows.
• Risk-Control Heatmap – color-coded risk mitigation strength.
• Outcome Impact Tree – linking KPI variance to applications.
• Data Lineage Lens – traces data from source to business outcome.

These visualizations are rendered through Power BI or the React NLQ UI for interactive exploration.

---

15 Takeaway #

The ontology is the operating system of EA 2.0.
Without it, AI has nothing to reason over; with it, every piece of data, every policy, every outcome becomes part of a single living model of the enterprise.