Every enterprise has a vocabulary problem.
The insurance division says "policy" and means a contract with a policyholder. The IT division says "policy" and means an access control rule. HR says "policy" and means an employee handbook section. When these three divisions share a database, a report, or a meeting room — the word "policy" creates confusion that no amount of context can fully resolve.
Now multiply that by every ambiguous term in the organization. "Account." "Case." "Service." "Agent." "Claim." "Asset." "Contract." Each word means different things to different people. And every time those people try to work together, they're speaking different languages with the same vocabulary.
The Upper Ontology
OACIS resolves this with formal ontology — specifically, with IEEE's Suggested Upper Merged Ontology (SUMO).
SUMO is the largest publicly available formal ontology in the world. It contains over 25,000 terms organized in a hierarchy that spans all domains of human knowledge — from abstract concepts like "Process" and "Relation" down to concrete ones like "FinancialTransaction" and "InsurancePolicy."
An ontology isn't a dictionary. A dictionary tells you what a word means in general. An ontology tells you what a concept is — its properties, its relationships, and its place in the structure of knowledge.
When OACIS maps the insurance division's "policy" to sumo:InsurancePolicy and IT's "policy" to sumo:Policy (a subclass of sumo:Proposition), the ambiguity disappears. Not through a naming convention or a glossary — through formal, machine-readable type classification.
The Layered Stack
SUMO sits at the top of the ontology stack. Below it:
ISO 21838 (BFO) — The Basic Formal Ontology, an ISO standard that provides the top-level structural categories: objects, processes, qualities, dispositions. It bridges SUMO's broad coverage to domain-specific precision.
Domain Ontologies — Industry-standard vocabularies that map to the upper layers: ACORD for insurance and financial services, HL7 FHIR for healthcare, ISO 20022 for financial messaging, CDM (Common Domain Model) for derivatives and securities.
Organization Taxonomy — The bottom layer: your company's specific terms, roles, products, processes, and entities. "Coastal Atlantic's Commercial Property Auto-Renew Process" maps upward through ACORD, through BFO, through SUMO — and the graph understands exactly what it is and how it relates to everything else.
Why Formal Ontology Beats Tagging
Most knowledge management systems use tags or categories. A document gets tagged "compliance" and "insurance" and "Q3." But tags are flat. They don't encode relationships. They don't resolve ambiguity. And they're maintained by humans who tag things inconsistently.
Formal ontology is different. It's not a flat list of labels — it's a hierarchical, relational structure where every term has:
- A type — what kind of thing it is
- Properties — what attributes it has
- Relationships — how it connects to other things
- Constraints — what rules govern it
- A position in the hierarchy — where it fits in the structure of knowledge
When the knowledge graph ingests a new piece of information, the ontology mapping doesn't just label it — it locates it in the structure of organizational knowledge. And once it's located, it's queryable, connectable, and computable.
Minimum Viable Taxonomy
The temptation with ontology projects is to try to model everything upfront. OACIS resists this with Principle Six: Ontological Mapping. Start with the smallest set of terms that resolves a real ambiguity or connects previously siloed knowledge. Grow through use, not through ambition.
SUMO provides the ceiling. Your organization's needs define the floor. The ontology grows to fill the space between them — one resolved ambiguity at a time.
This post draws from Chapter 6: The DevOps of Everything and Chapter 8: The Knowledge Graph of Organizations as Code: The Intelligent System Revolution.
← Back to Blog