Ontology

The foundational building blocks of an ontology. Classes define the abstract categories of things that exist in the manufacturing domain, such as Machine, Material, Sensor, or WorkOrder. Instances (or individuals) are the concrete, specific members of those classes, such as CNC-Lathe-42 (an instance of Machine) or Steel-Grade-316L-Batch-7 (an instance of Material). A well-defined class hierarchy ensures that all data is consistently categorized.

• Class: A blueprint or set definition (e.g., Pump).
• Instance: A specific real-world entity (e.g., Pump-23 on Line 4).
• Subclass: A more specific category inheriting parent properties (e.g., CentrifugalPump is a subclass of Pump).

Properties define the attributes of classes and the relationships between instances. Object Properties link one instance to another, forming the edges of the knowledge graph. For example, the property hasPart connects an Engine instance to a Piston instance. Data Properties link an instance to a literal value, such as a string, number, or date. For example, a Machine instance has a data property serialNumber with the value "CNC-2023-0042".

• Object Property: isMaintainedBy (links Machine to Technician).
• Data Property: maxOperatingTemperature (links Machine to 150^^xsd:integer).
• Domain and Range: Constraints that specify which classes can use a property and what type of value is expected.

Axioms are the logical rules and constraints that enable automated reasoning. They move an ontology from a simple data model to a machine-interpretable logic system. Restrictions define anonymous class expressions that constrain property values. For instance, a CriticalAsset can be defined as any Machine that hasFailureMode at least one CatastrophicFailure. A reasoner can then automatically classify any machine meeting this criterion.

• Existential Restriction: A Pump must have at least one Impeller (e.g., hasPart some Impeller).
• Cardinality Restriction: A Car has exactly four Wheel instances.
• Disjointness Axiom: A Sensor cannot simultaneously be an Actuator.

The backbone of any ontology is its taxonomic, or 'is-a', hierarchy. This is a tree-like structure where specific concepts are linked to more general ones via a parent-child relationship. This enables inheritance: a CentrifugalPump automatically inherits all properties and constraints defined for Pump, which in turn inherits from RotatingEquipment. This structure is critical for querying at the right level of granularity, such as finding all RotatingEquipment due for maintenance, regardless of its specific subtype.

• Generalization: VibrationSensor is-a Sensor is-a Device.
• Inheritance: If Device has a manufacturer property, VibrationSensor inherits it.
• Polyhierarchy: A concept can have multiple parents (e.g., a Robot is-a Machine and is-a AutomatedSystem).

All knowledge in an ontology is ultimately expressed as semantic triples: subject-predicate-object statements that encode a single, unambiguous fact. This is the native data structure of RDF and the lingua franca of the semantic web. A triple like Pump-23 hasFailureMode BearingFatigue is a self-contained fact that can be queried, linked, and reasoned over. The power of the ontology lies in the network effect of millions of these interconnected triples.

• Subject: The entity being described (e.g., Pump-23).
• Predicate: The property or relationship (e.g., hasFailureMode).
• Object: The value or related entity (e.g., BearingFatigue).
• Graph Pattern: Pump-23 -> hasPart -> Bearing-7 -> exhibits -> HighFrequencyVibration.

To ensure global uniqueness and prevent naming collisions, every class, property, and instance in a formal ontology is identified by a Uniform Resource Identifier (URI). These URIs are typically organized into namespaces, which act as prefixes for a set of related terms. For example, a standard ontology like Dublin Core uses the namespace dc, so its creator property is dc:creator. A company might define its own namespace, acme, for proprietary concepts like acme:CNCLathe.

• Global Uniqueness: http://acme.com/ontology/Pump is distinct from http://example.org/Pump.
• Prefix: A shorthand for a namespace URI (e.g., acme: for http://acme.com/ontology/).
• Standard Namespaces: rdf:, rdfs:, owl:, xsd: provide core modeling primitives.

A hierarchical classification scheme that organizes concepts into parent-child relationships. While an ontology defines the nature of relationships (e.g., 'causes', 'requires'), a taxonomy provides the controlled vocabulary and 'is-a' inheritance tree. In manufacturing, a taxonomy might classify a 'Centrifugal Pump' as a child of 'Rotating Equipment', which is a child of 'Physical Asset', forming the backbone upon which richer ontological rules are layered.

A semantic web language designed for rich, complex knowledge representation. OWL provides greater machine-interpretability than RDF by enabling formal logic-based reasoning. It allows you to define cardinality constraints (e.g., a Machine has exactly one Operator), transitive properties, and class equivalence, allowing a reasoner to infer new facts from explicitly asserted data in a manufacturing ontology.

A software component that applies logical inference rules to an ontology to derive new, implicit facts. For example, if an ontology states 'BearingFatigue isA FailureMode' and 'HighFrequencyVibration indicates FailureMode', a reasoner can automatically classify a newly observed vibration pattern as a specific fault type without explicit programming. This is the engine that makes ontologies actionable.

A W3C standard for validating RDF graphs against a set of conditions. SHACL ensures that manufacturing knowledge graph data conforms to the expected shapes, cardinalities, and data types defined by the ontology before being used for critical analysis. It acts as a schema guardrail, preventing nonsensical assertions like a Pump having a negative operating temperature.

The ability of two or more systems to exchange information and have the meaning accurately interpreted by the receiving system. This is the primary goal of an ontology. By committing to a shared, formal ontology, a SCADA system and an ERP system can agree that 'DowntimeEvent-42' refers to the same unplanned stoppage, enabling true cross-system automation without brittle point-to-point translations.

An international standard defining a hierarchical model of manufacturing operations (Level 0 through Level 4). It serves as a canonical, pre-built ontology for integrating business and production data. Mapping proprietary machine data to ISA-95's formal concepts of 'Work Center', 'Production Schedule', and 'Material Lot' provides immediate semantic alignment across the enterprise.