semantic model

A semantic model is a structured representation of the meaning of data, concepts, and their relationships, designed so that different systems and stakeholders interpret information in a consistent way. It focuses on what data represents in the real world, not just how it is stored or formatted.

What a semantic model includes

In industrial and manufacturing contexts, a semantic model commonly defines:

• Business and operational concepts, such as equipment, batches, shifts, work orders, lots, or KPIs like OEE or first-pass yield.
• Attributes of those concepts, such as units, status, time intervals, and calculation windows.
• Relationships between concepts, such as which machines belong to which line, or which events contribute to a particular KPI.
• Constraints and rules, such as how a metric is calculated, required dimensions, or allowed value ranges.

The goal is to provide a shared vocabulary and structure so that terms like “Downtime”, “Batch”, or “Scrap” have clearly defined and traceable meanings across MES, historians, quality systems, ERP, and partner systems.

How semantic models are used operationally

Operationally, semantic models are used to:

• Map heterogeneous data sources (for example, different plant-specific tags and tables) into a shared conceptual view.
• Harmonize KPIs and metrics so that comparisons across sites, lines, or partners are based on compatible definitions.
• Support integration between OT and IT systems by providing a common layer between physical tags, database schemas, and business applications.
• Enable traceability of meaning, by documenting calculation rules, versions, and the origin of definitions.

A semantic model does not need to replace existing systems or databases. It can sit on top of them as a logical layer that interprets and standardizes their data.

Relation to data models and ontologies

A semantic model is related to, but distinct from:

• Physical or logical data models, which describe how data is stored (tables, columns, tags, message schemas). A semantic model focuses on meaning, not storage details.
• Ontologies and taxonomies, which are formal or hierarchical structures of concepts. Many semantic models in industry are practical ontologies, but they may be less formal than those used in academic knowledge representation.

In manufacturing, semantic models are often aligned informally with standards such as ISA-95 or sector-specific reference models, without necessarily replicating them in full.

Common confusion

• Semantic model vs. KPI catalog: A KPI catalog lists metrics and basic formulas. A semantic model goes further by defining the underlying concepts, relationships, and conditions so the same KPI can be computed consistently across contexts.
• Semantic model vs. master data: Master data defines specific entities and values (for example, material codes, customer IDs). A semantic model defines the meaning and structure that master data instances follow.
• Semantic model vs. integration mappings: Point-to-point mappings convert fields between systems. A semantic model provides a shared target meaning that those mappings align to, reducing system-specific coupling.

Context: harmonizing KPI semantics across plants and partners

When used as a bridge across plants and external partners, a semantic model provides a shared definition layer for KPIs and operational concepts. Local systems can keep their own tag names, table structures, and calculation specifics, while mappings connect them to the central semantic model. Versioning and traceability within the model help document how definitions evolve over time and how each site or partner maps to the shared semantics.