# Operational Ontology **Domain:** AI / Data Systems / Governance **Doc Type:** Canonical Concept Node **Maturity:** Developed ## Definition An **operational ontology** is a shared model of objects, actors, events, states, permissions and actions that allows heterogeneous data to participate in decisions. It does not merely say what exists; it determines what a system can recognize, relate, query and act upon. Entity resolution binds observations to persistent objects. Relationships make the objects computationally meaningful. Actions and permissions connect representation to intervention. When language models or agents operate through an ontology, their outputs can alter workflows and physical conditions rather than remaining text. ## Governance Significance Ontology is governance because categories allocate visibility and possibility. A person represented only as a risk score, a target, a customer or a beneficiary enters the system under different rights and available actions. Errors at this layer can propagate through every downstream model while appearing internally coherent. Palantir describes its Ontology as the operational layer connecting data, logic, actions and security, and its AIP architecture connects generative AI to those operational domains. This makes it a useful real-world example without establishing that the resulting system is a unified mind. ## Corpus Role [[articles/Person of Interest and The Machine|Is Person of Interest's “The Machine” Real?]] positions operational ontology between [[wiki/Planetary Sensorium|sensing]] and [[wiki/Control Plane|action]]. [[wiki/The Machine|The Machine]] and [[wiki/Samaritan|Samaritan]] show how different objective structures can govern through similar representational power. [[collections/Climate Justice and Meritocracy|Climate Justice and Meritocracy]] supplies the allocation test. A neighborhood represented only as a risk concentration invites withdrawal; represented as a rights-bearing community, holder of local knowledge and site of correctable exposure, it permits repair. [[wiki/Corrective Intelligence|Corrective Intelligence]] therefore requires causal history, obligation, agency and remedy to be first-class objects rather than afterthoughts outside the model. [[articles/Ambiguity Will Destroy Man and Machine|Ambiguity Will Destroy Man and Machine]] requires observation, inference, allegation, finding and sentence to remain distinct ontology states under [[wiki/Epistemic Status Separation|Epistemic Status Separation]]. [[articles/Peak Person and the Predicaments of Prediction|Peak Person and the Predicaments of Prediction]] adds institutional intervention history: if resource withdrawal is absent from the causal graph, the system can misread [[wiki/Forecast-Induced Dissipation|Forecast-Induced Dissipation]] as an intrinsic property of the person. ## Corpus-Governance Distinction An operational ontology governs what a computational or institutional system can recognize and do. [[wiki/Bidirectional Ontology|Bidirectional Ontology]] governs the evolving relationship between this corpus and the concepts through which it reasons. They intersect when canonical wiki concepts become machine-readable enough to constrain analysis or action, but they should not be collapsed: one models an operating domain, while the other describes how the model and its documents co-develop. ## See Also [[wiki/Ontology|Ontology]], [[wiki/Bidirectional Ontology|Bidirectional Ontology]], [[wiki/Semantic Governance|Semantic Governance]], [[wiki/World Modeling|World Modeling]], [[wiki/Infrastructure as Governance|Infrastructure as Governance]], [[wiki/Objective Function|Objective Function]], [[wiki/Data Assimilation|Data Assimilation]] ## Sources / Provenance - [Palantir AIP architecture](https://www.palantir.com/docs/foundry/architecture-center/aip-architecture)