At the top-most level, the concepts within the theory can be seen as being represented in a standard frame-slot-filler style of notation. However, the underlying representation is a conceptual graph--every frame, every slot, and every filler exists in the knowledge base as specific nodes in the graph. Slots are nodes which act as relationships between other nodes within the knowledge system. In addition to the actual representation, there are certain elements which all concepts share. First, all concepts possess a current function. This represents which known role the concepts is filling at the current point in reasoning. Second, supporting this is the idea that certain slots represent primary attributes--those features of the concept which allow it to accomplish its function. Finally, the remaining slots represent secondary attributes--features of the concept which are known but which are not currently contributing to the explanation of how it achieves its function.
This representational commitment allows me to precisely define what ``novel'' means with respect to understanding of unknown concepts. Novelty must always be judged from the perspective of a particular reasoner--what is novel to one person may be mundane to another. With that in mind, think about a black longsword. If you then consider a red longsword, you are reasoning about a concept I call instantiationally novel (I-Novel). The function is the same, the primary attributes are the same, but a secondary attribute has been altered. Now, consider a shortsword. While the function has remained the same, the primary attributes have changed. This concept possesses Evolutionary Novelty (E-Novelty). Finally, think for a moment about a light saber from Star Wars. The function remains the same but the primary and secondary attributes have been completely altered. This is an example of a concept which possesses Revolution Novelty (R-Novelty) with respect to the original concept, the black longsword.
Beyond the basic representation, my research makes a specific claim about how the knowledge is organized. This is necessary since I am dealing with the understanding of novel concepts--for the creative understanding algorithm to correctly handle unknown concepts, it attempts to place them into an organization of known concepts. So, the concepts which are in the system must exist in an organization which supports this. In particular, the knowledge representation system underlying the creative understanding theory organizes all concepts into one of twenty high-level ontological categories. Five domains of knowledge have to be represented: physical, mental, social, emotional, and temporal. Also, four types of knowledge are represented: objects, agents, actions, and states. To achieve the twenty basic categories, one simply crosses these two dimensions of knowledge. Finally, a small set of heuristics exist which describe the permissible ways that concepts may transition around the ontology. For example, concepts may transition in such a way that they are left in the same category. More complex transitions would move the concepts into other categories; for instance, an object may transition to an action by creating an action which captures a function of that object.