· 3 min read

Designing a Threat Ontology

This ontology can be used in threat modeling. v1.0

A TARA (Threat Analysis and Risk Assessment) is a structured methodology for identifying threats to a system, assessing their likelihood and impact, and determining appropriate controls. It is used across regulated industries — medical devices, automotive, aviation, industrial control systems — wherever cybersecurity risk must be formally documented and accepted.

TARAs are inherently graph-structured. Actors target assets. Attacks decompose into steps. Risks are mitigated by controls. Decisions cite authority. This relational structure maps directly to RDF, enabling TARA data to be stored as a knowledge graph, queried with SPARQL, validated with SHACL, and reasoned over with OWL.

The ontology above defines the classes and properties that make this possible.

Classes

ClassDescription
tara:ActorAny entity capable of initiating a threat. Superclass of tara:ThreatActor.
tara:ThreatActorrdfs:subClassOf tara:Actor. A typed actor characterized by capability and intent.
tara:CapabilityWhat a threat actor can technically execute — exploit development, physical access, supply chain insertion.
tara:IntentMotivation — financial, geopolitical, competitive, ideological. Both capability and intent must be present for a credible threat.
tara:AssetAnything of value that can be threatened: firmware, cryptographic keys, sensor data, a communication channel, a physical interface.
tara:AttackAn ordered sequence of steps that, if completed, results in a realized threat against one or more assets.
tara:AttackStepAn atomic action within an attack. Each step uses a technique and may target an asset directly.
tara:TechniqueA specific method of execution, drawn from a taxonomy such as MITRE ATT&CK, ICS-ATT&CK, or a domain-specific equivalent.
tara:RiskThe probability-impact pair resulting from a successful attack. The central object around which mitigation and acceptance are organized.
tara:OutcomeThe real-world consequence of a realized risk: patient harm, service disruption, data exfiltration, regulatory penalty.
tara:ControlA countermeasure — preventive, detective, or responsive — that reduces risk to an acceptable residual level.
tara:DecisionThe formal risk treatment decision: mitigate, accept, transfer, or avoid.
tara:JustificationThe rationale for a risk acceptance decision. Must be traceable and auditable.
tara:AuthorityThe standard, regulation, or guidance that grounds a justification: ISO 21434, IEC 62443, FDA cybersecurity guidance, NIST SP 800-53.

Properties

PropertyDomainRangeDescription
tara:targetstara:Actortara:AssetAn actor has selected an asset as an objective.
tara:executestara:ThreatActortara:AttackLinks an actor to the attack chain they carry out.
tara:hasCapabilitytara:ThreatActortara:CapabilityAsserts the technical means available to the actor.
tara:hasIntenttara:ThreatActortara:IntentAsserts the motivation driving the actor.
tara:containstara:Attacktara:AttackStepStructural composition — an attack is fully described by its steps.
tara:precedestara:AttackSteptara:AttackStepEstablishes execution order within an attack chain.
tara:targetstara:AttackSteptara:AssetA specific step acts directly against an asset.
tara:usestara:AttackSteptara:TechniqueMaps each step to its technique, enabling cross-product pattern analysis.
tara:producestara:Attacktara:RiskA successful attack produces a risk with defined likelihood and severity.
tara:leadsTotara:Risktara:OutcomeConnects the abstract risk to its concrete consequence.
tara:mitigatestara:Controltara:RiskThe core mitigation relationship.
tara:selectstara:Decisiontara:ControlA decision selects one or more controls as the treatment.
tara:acceptstara:Decisiontara:RiskA decision formally accepts a residual risk after controls are applied.
tara:requirestara:Decisiontara:JustificationEvery decision requires a justification.
tara:citestara:Justificationtara:AuthorityA justification cites the authority that makes acceptance defensible in audit or regulatory review.

TARA Threat Ontology v1.0 — February 20, 2026 — 3 min read