Many people treat concept maps, knowledge systems, knowledge graphs as interchangeable or assume that the difference is simply one of scale, i.e. “small vs large”. That framing, however, misses the real distinction.
The difference between Concept Maps and Knowledge Systems is not one of scale, but of purpose and use case: One helps us understand, while the other helps us organize and sustain knowledge. We will see that they complement each other, while they have different purposes and emphasis.
What is a Concept Map?
A concept map is a visual representation of concepts and their relationships.
Concept Maps render a set of nodes (concepts) and labeled links (relationships) between the concepts. They are often centered around a guiding theme, the focused node. This visualisation helps to understand by making the neighborhood, the surrounding concepts visually explicit, close to the eye. For instance, the weekday “Thursday” is defined by its neighbors “Wednesday” and “Friday”. Such maps help to explore – like a real traffic map. Thus they are great for learning, teaching, and understanding. They visualize knowledge and make it explicit particularly for humans.
- Visualise and Explore
- Zoom in: Learn the details
- Understand: What is the meaning of a concept, a node
- Size: Small data collections
- Known from: Mindmaps, Visualised hyperlinks
However, they are difficult to scale, both from a technical as well as a usability perspective. Since a single concept can have many relations to other concepts, the amount of relations, i.e. the lines on the map, quickly multiplies. With just 15 concepts on a map, each with a couple of relations … you can end up easily with 100 lines on the screen. While this may look impressive in screenshots, we believe it has a serious usability flaw. It is too “noisy”.
What is a Knowledge System?
Theoretic publications often define it somewhat like this: “a structured system for organizing, storing, and retrieving knowledge across a domain”. In a slightly more practical terms, we may say: “a systematic inventory of the elements that a domain consists of”. Taxonomies and ontologies are very typical examples of knowledge systems.
- Organize and control
- Zoom in and out: Where is the right spot?
- For humans and machines
- Size: Large(r) database
- Known from: Taxonomies, Nomenclatures, larger ontologies.
Defined relationships, designed for re-use and scalability across large domains. They enable consistency and are often intended to support machines (search, AI, automation). Their purpose and value lies in managing a whole domain. It reflects a so-called Information Architecture.
Comparison
Concept Maps
- Exploring and learning
- Informal
- Smaller amount of nodes
- Primarily for humans
- Easy to change
Knowledge Systems
- Organization and retrieval
- Formal
- Scale 1000s of nodes
- For humans and machines
- Controlled evolution
Thus, concept maps and knowledge systems are not competitors. They often simply represent different stages of evolution, maturity, or completeness. While concept maps are close to what one might call “sketching”, knowledge systems act like an architectural blueprint of a domain.
The takeaway? Use maps for learning, brainstorming, explaining ideas, and illustrating the meaning of a concept. Use knowledge systems for managing larger volumes, for supporting Search or AI.
Concept maps help us figure things out.
Knowledge systems help us keep things organized once we have.
And where does then Knowledge Graph fit in here?
In short we could say that a Knowledge Graph is a technical representation of a knowledge system.
It is a structured representation of entities and their relationships, typically modeled as a graph (nodes + edges), often with formal semantics. Thus a knowledge graph is about storage: machine-readable predicates over a potentially massive amount of data, i.e. the nodes in the graph.
Important: We understand this as a graph about “knowledge”, not any graph of facts.
Summary and Outlook
A concept map shows how you think things relate. A knowledge system organizes a whole domain. And a knowledge graph encodes relationships so that machines can reason over them. Concept maps can inspire knowledge systems, whereas knowledge graphs can formalize the data. A concept map, a knowledge system – they become a knowledge graph only when their relationships are formalized, standardized, and made machine-readable. With this you feed explicit, machine-processable knowledge into AI applications, make them trustworthy.
- Concept Map → explore ideas
- Knowledge System → organize knowledge
- Knowledge Graph → compute and scale knowledge relationships
In Coreon we have used the term Concept Map for one of the UI widgets for many years. This widget renders the relations between concepts. However, since Coreon not only illustrates the neighborhood of a concept but also organizes and controls large amounts of concepts in a taxonomic, hierarchical fashion we are considering to rename this widget in the new Coreon 2. The discussion is still ongoing internally and with the beta community. Candidates include “Visual Knowledge Graph”, “Hierarchical Map”, or simply “Graph” … we’ll see :-).
