Concept Models and Ontologies
The Sketch and The Blueprint
The Distinction Between Concept Models and Ontologies
In the knowledge representation and information science domains, the terms concept model and ontology frequently intersect, yet they denote fundamentally different constructs with distinct purposes, formalisms, and applications. Understanding their differences is crucial for effective knowledge management, system design, and trans-disciplinary communication. This essay explores the essential distinctions between these two data modeling methodologies , elucidating their unique roles in organizing and representing knowledge.
Defining the Terms
A concept model is a conceptual representation of ideas, entities, and their relationships within a specific domain or context. It serves as a cognitive map that helps stakeholders visualize and communicate abstract structures, typically through diagrams, narratives, or informal specifications.
Concept models prioritize clarity and shared understanding over formal rigor, making them accessible to domain experts without technical backgrounds. They answer the question: “What ideas exist in this domain, and how do they relate to each other?” For instance, a concept model for a healthcare system might illustrate relationships between “Patient,” “Treatment,” and “Outcome” without specifying exact computational rules.
An ontology, by contrast, is a formal, explicit specification of a shared conceptualization. It is a rigorous, machine-readable framework that defines the types, properties, and interrelationships of entities in a domain through precise logical axioms. An ontology is a logical reasoning model, by way of adherence to the encoded rules and obligations.
More About Ontology
Note that there is a distinction between ontology ala the philosophy domain and ontology as defined by information science. The two distinct concepts share common ground, in their focus on concepts and categorization. Information science ontologies are not generally open deliberations but more deterministic, by nature. If we were stuck pondering the existence of a thing as is the case in philosophy’s brand of ontology, we would never build an information science oriented ontology, because we would be arbitrating the aboutness of a thing and stuck in the dialogue of what isness.
Ontologies can be expressed formally, using OWL (Web Ontology Language). To be clear, OWL itself is not an ontology, it’s a way to logically represent and describe knowledge. There are different flavors of OWL, with different capabilities and depths of semantic expression. Ontologies also utilize Resource Descriptive Framework (RDF), a microformat and standard data model, used to describe things and their relations to one another.
Ontologies enable automated reasoning, inference, and semantic interoperability by way of its rule bases, construct and framework. OWL and RDF are tools and formats for modeling and representing knowledge but they are not ontologies. This distinction is critical, to understand the world of ontologies.
An ontology is constructed of classes, properties, attributes and relations. Ontologies are generally constructed and modeled to represent topics, domains, themes, workflows, entities, products, applications, temporal data, spatial data—the list goes on. You can shop for open standard ontologies on the web and mix and match only what you need, when building an ontology.
‼️ NOTE: **I will be publishing a reference document shortly, containing a growing list of W3C standard ontologies, categorized according to use, domain and type.**
Put Your Finger on the Thread
Ontologies answer the question: “What can be logically proven about these entities, and how can machines consistently interpret them?” In a healthcare domain, an ontology would formally define that a “Patient” is a subclass of “Person,” that “hasTreatment” is a functional property, and that certain relationships entail specific logical constraints.
To build an ontology, ontologists generally use competency questions, as a means of defining and scoping what is to be included and excluded in any particular ontology modeling exercise. Competency questions are a human-in-the-loop (HITL) necessity, and are conducted in natural language, and are normally part of the ontology design process.
