Data Architecture
The metadata layer that makes sovereign intelligence possible.
> node_id: DOC-2024-0892-S4.2
> source: Annual Report 2024
> confidence: 0.97
> provenance: VERIFIED
> topic: supply_routes
The Foundation
A structural approach to content where every piece of data carries its own metadata.
Every node tagged with: source, confidence rating, topic, provenance score, timestamp, relationship map. This is not metadata bolted on after the fact. OBM is the architecture itself. The data and its context are inseparable.
Traditional systems store content in one place and metadata in another - if they store metadata at all. OBM inverts this. Every atomic object knows what it is, where it came from, how reliable it is, and what it connects to.
The document is the database.
The Process
From raw content to structured intelligence in five stages.
> Raw content ingested
Documents, recordings, archives, databases - all formats accepted.
> Content decomposed into atomic objects
Each content node broken into the smallest meaningful unit.
> Every object tagged with full OBM schema
Source, confidence, provenance, topic, relationships - all assigned.
> Objects linked in a queryable graph
Relationships mapped across the entire corpus. Every connection explicit.
> Archive becomes structured intelligence
A searchable, queryable system built from your own verified data.
Every claim traces to its origin document. No ambiguity about where a piece of information came from. The citation is built into the object, not appended afterwards.
Know how reliable each piece of information is. Every node carries a confidence rating so you can weigh outputs based on the quality of the underlying data.
See how concepts connect across your entire corpus. OBM makes implicit relationships explicit, turning isolated documents into a connected intelligence graph.
AI queries hit the right data, not all of it. Structured metadata means the model retrieves only what it needs, reducing token cost and improving answer quality.
If the model cannot find a verified source, it says so. OBM enforces a hard boundary between what is known and what is not. No fabricated citations. No confident-sounding nonsense. Every answer either has provenance or it declares uncertainty.
The Standard
Every object in the system conforms to this schema. No exceptions. This is what makes the data queryable, verifiable, and sovereign.
The schema is designed to be extended per domain - additional fields for legal, medical, defense, or media contexts - but these core fields are always present.
> node_id: DOC-2024-0892-S4.2
> type: text_segment
> source: Annual Report 2024, Section 4.2
> confidence: 0.97
> provenance: VERIFIED
> topic: ["supply_routes", "eastern_corridor"]
> relationships: [DOC-2024-0891, DOC-2024-0893]
> timestamp: 2024-03-15T09:22:00Z
node_id uniquely identifies every atomic object in the system
confidence scores from 0.0 to 1.0 rate data reliability
provenance tracks verification status: VERIFIED, PENDING, UNVERIFIED
relationships link this node to every related object in the graph
Where OBM Lives
OBM is not a standalone product. It is the foundational layer that runs through every vertical Third ARK offers. AI Ready designs your OBM schemas. Sovereign Intelligence queries them. The Content Engine produces cited output from them.
Without OBM, there is no citation. Without citation, there is no sovereignty.
Every sovereign system starts with the right architecture.
enquiries@third-ark.com