ΟΝΤΟΛΟΓΊΑ
The Study of Being — from Parmenides to Palantir
A study guide to the philosophical, academic/technical, and popular uses of one of the most overloaded words in the English language.
Begin Reading ↓Roadmap
"Ontology" means fundamentally different things depending on who is speaking. This guide tracks it across three domains.
The original sense: metaphysics, the study of being & existence, 2,500 years of debate.
02The engineered sense: formal, machine-readable models of knowledge for reasoning systems.
03The borrowed sense: business buzzword, social theory, anthropology's "ontological turn."
Part I
The study of being, existence, and the basic categories of reality
Philosophy — Origins
ὄν / ὄντος (on / ontos)
"being" — present participle of einai, "to be"
+
-λογία (-logia)
"study / discourse of"
Literally: "the study / science of being"
The ancient Greeks originated the inquiry but never used this word. "Ontologia" first appears in Latin in the early 1600s.
Philosophy — Core Questions
What does it mean "to be"? Is existence itself a property, or presupposed by having any property at all?
What is the totality of things that exist — concrete objects, abstract objects, possible or fictional entities?
How should basic kinds of being be classified? (Aristotle: substance, quantity, quality, relation…)
Do general properties (redness, humanity) exist independently of the things that instantiate them?
Is there a fundamental substratum underlying properties, or is reality just bundles of relations?
What makes an entity the same thing over time? How do essence and existence relate?
Philosophy — History
"Father of ontology." What-is (to on) is one, eternal, unchanging; what-is-not cannot be thought.
Categories & Metaphysics: being qua being; substance (ousia) as primary reality; 10 categories of being.
Abelard (conceptualism), Aquinas (essence/existence), Duns Scotus (univocity of being, haecceity), Ockham (nominalism, Ockham's Razor).
Substance dualism: res cogitans (mind) and res extensa (matter) as two really distinct substances.
Spinoza: one substance ("God or Nature"). Leibniz: infinite immaterial "monads" in pre-established harmony.
Decisive critique of dogmatic ontology: we cannot know things-in-themselves; ontology becomes a study of the categories that structure experience.
Science of Logic: being and thought are identical; categories unfold dialectically toward Absolute Spirit.
Being and Time launches "fundamental ontology" — the forgotten question of Being, explored via Dasein's being-in-the-world.
Being and Nothingness: being-in-itself vs. being-for-itself; "existence precedes essence."
"On What There Is": "to be is to be the value of a bound variable" — ontological commitment via logic, founding analytic metaontology.
Philosophy — Schools
Do universals (properties, kinds) exist mind-independently, are they mere names, or only concepts? The classical "problem of universals."
Is reality fundamentally physical, fundamentally mental, or composed of two irreducible substances?
Reality is constituted by events and becoming, not static substances (Whitehead, Heraclitean flux) — opposed to "substance ontology."
How do collective intentionality and agreement bring institutions, money, and nations into being? (Searle, Epstein, Thomasson)
Philosophy — Live Issues
Is "what exists?" a single well-posed question (Quine, van Inwagen) or merely verbal/framework-relative (Carnap, Thomasson's "easy ontology")?
Include only fundamental, "joint-carving" entities, or a generous inventory of composites, artifacts, and abstracta?
Beyond what exists — what grounds what? Reviving Aristotelian priority in a new idiom (Schaffer, Fine).
Do numbers, propositions, and possible worlds really exist (Platonism), or are they useful fictions?
How much of race, gender, money, and borders is constructed yet still objectively real?
Endurantism (wholly present at each moment) vs. perdurantism (temporal parts / four-dimensionalism).
Part II
Formal, machine-readable models of knowledge — borrowed from philosophy in the 1990s
CS/AI — Definition
"An ontology is a formal, explicit specification of a shared conceptualization." Gruber (1993) → Borst (1997) → Studer, Benjamins & Fensel (1998)
An abstract, simplified model of some phenomenon in the world.
Types of concepts & constraints are stated openly, not left implicit in code.
Machine-readable / machine-interpretable, expressed in a logic-based language.
Captures consensual knowledge accepted by a community, not one person's view.
In practice: a controlled, structured vocabulary of classes and relations plus formal axioms that constrain how those terms combine — enabling automated reasoning, consistency checking, and knowledge sharing between systems.
CS/AI — Anatomy
Categories of things (e.g. Person, Disease) organized into is-a hierarchies.
Concrete instances of classes — e.g. "Marie Curie" is-a Person.
Typed links between individuals/classes (hasChild, partOf, locatedIn).
Links from individuals to literal values (hasAge → integer).
Logical constraints enabling inference — disjointness, necessary & sufficient conditions.
The subclass "is-a" tree — the taxonomic skeleton extended by relations & axioms.
CS/AI — Standards
W3C standards, from raw data up to logical reasoning
CS/AI — In Practice
DBpedia (Wikipedia → RDF triples) and Wikidata power the "Linked Open Data cloud."
Gene Ontology (gene product attributes) and SNOMED CT (clinical terminology in EHRs worldwide).
Google Knowledge Graph and schema.org markup let search engines disambiguate entities.
Product taxonomies, customer models, and supply-chain semantics for data integration & master data management.
Protégé (Stanford) is the standard open-source editor; reasoners like HermiT and Pellet check consistency.
SUMO, DOLCE, and BFO aim to be domain-independent, reusable foundations across fields.
CS/AI — Comparison
| Artifact | Structure | Semantics | Typical Use |
|---|---|---|---|
| Taxonomy | Single tree (is-a / part-of) | Informal, no axioms | Categorization, navigation |
| Thesaurus | Hierarchy + broader/narrower/related | Informal, library-science conventions | Controlled vocab for indexing |
| Ontology | Classes + relations + formal axioms | Formal, machine-interpretable, enables inference | Knowledge representation, reasoning, AI |
| Data Model | Entities, attributes, relationships | Structural, application-bound, closed-world | Application / database design |
Rule of thumb: glossary → controlled vocabulary → taxonomy → thesaurus → data model → formal ontology — increasing formality & inferential power at each step.
CS/AI — In the LLM Era
Ontology-backed knowledge graphs ground retrieval-augmented generation, reducing hallucination and enabling multi-hop, explainable reasoning.
Unique URIs let AI systems resolve ambiguous mentions ("Jaguar" the animal vs. the car) — something plain embeddings struggle with.
Ontologies supply a symbolic backbone — types, hierarchies, constraints — to validate or constrain generative output.
Companies pair LLMs with domain ontologies so AI assistants query governed structured data, improving accuracy and auditability.
Continuity: today's "knowledge graphs" are directly descended from 1970s semantic networks and 1990s ontology engineering — the same representational paradigm, now infrastructure for grounded, verifiable generative AI.
Part III
How "ontology" escaped philosophy — into business, social theory, and anthropology
Popular — Business
In data, marketing, and enterprise-tech contexts, "ontology" is frequently flattened to "a structured way of organizing things and their relationships" — richer than a taxonomy, but far from Aristotle's original stakes.
A model of entities, attributes & relationships in a business domain — relational/web-like vs. a taxonomy's strict tree.
Marketing agencies use it for a brand's identity, purpose & voice — essentially a strategy document with borrowed prestige.
The most prominent commercial actor: "the Ontology" maps datasets to object types & actions, building an operational "digital twin" of an organization.
Popular — Other Disciplines
Searle's Construction of Social Reality: money, marriage & nations are "institutional facts" ("X counts as Y in context C").
"Ontological security theory" (Mitzen, Steele): states seek a stable sense of identity, not just physical safety.
Ontology maps rich typed relationships (habitat, status) beyond a taxonomy's strict parent-child hierarchy.
Judith Butler's "relational social ontology" grounds vulnerability & interdependency as basic to human being.
Viveiros de Castro's "perspectivism" and Descola's four modes of nature/culture: societies inhabit genuinely different realities, not just different beliefs ("multiple ontologies").
Popular — Pitfalls
Ontology asks what exists / the nature of being. Epistemology asks how we know. The single most common conflation in business & social-science writing.
Tech/business usage often flattens ontology to any structured taxonomy or schema — losing the sense of a theory about what fundamentally exists.
Brand consultants and startup pitches invoke "ontology" (and cite Aristotle) to lend intellectual weight to what is functionally a categorization framework.
Many assume ontology just means a strict tree. Its defining feature — in both philosophy and CS — is capturing non-hierarchical relations too.
Synthesis
| Philosophy | Computer Science / AI | Popular / Cross-Disciplinary | |
|---|---|---|---|
| Asks | What exists, fundamentally? | How do we formally model a domain? | How do we organize / relate to reality? |
| Method | Argument, logical analysis | Formal logic, ontology languages (OWL) | Ethnography, business practice, theory |
| Output | Theories of being & categories | Machine-readable class/relation graphs | Frameworks, cultural claims, buzzwords |
| Key name | Aristotle, Heidegger, Quine | Gruber, Berners-Lee, Protégé | Searle, Viveiros de Castro, Palantir |