23 June 2005

Components of Ontology: Concepts

In paper 05E-SIW-045, the idea of a formal ontology was defined to comprise four different components. These are concepts, entities, relations, and rules. Here are a couple of ideas I'm having about concepts, based on the original paper and also based on a conversation held today with my thesis advisor - Andreas.

Concepts, in 05E-SIW-045, are defined as anything within the universe of discourse with addressable properties. This definition was chosen deliberately, for several reasons.

The first reason for defining concept as anything with addressable properties is based on the concept of a fractal ontology. In a fractal ontology, each layered view of the contents of the ontology can be redefined at a higher resolution view (where more details) or a lower resolution view (with fewer details). When you consider the lowest resolution that you can consider an ontology in - which is perhaps just the idea of "concepts" as a class, and then move through all the various layers of redefinition (starting with the first layer, where concepts are divided up into objects, events, actions, and phenomenology), each layer has more concepts, but each of these concepts is defineable by higher-resolution, comprising concepts. From the perspective of a data model based on our ontology, the highest resolution concepts would be property-values. These are enumerated details defining (and indeed, are instances of) properties. Properties, in turn, are then collected together into propertied-concepts. Propertied-concepts are higher-order ideas, things or entities, that are an abstracted way of thinking of the sum of the comprising propertied-concepts. These are then collected, at a level even further removed from the highest resolution property-vales, into associated-concepts. Associated-concepts are again collections of various propertied-concepts into higher order ideas (which, when considered apart from their comprising propertied-concepts, are necessarily at a lower resolution). And so on up the spiral away from resolution, but more towards higher and higher ordered ideas.

The second reason for defining concept as anything with addressable properties, is because they must needs be considered seperately, and individually at the more higher-resolution levels, due to the fact that they have to be re-assemblable into more than one higher order idea. Much as atoms are able to be combined with other atoms to make a whole bewildering set of different molecules, so the concepts of our ontology must be combined with other concepts to create a large number of other higher order constructions. Consider the battlesphere domain. Here we have a concept known as fuel (a gross simplification). It, itself, is a combination of other higher resolution concepts (the component elements, perhaps, the type of fuel, it's grade, etc). Now fuel is a concept that is combined with other concepts to create a number of higher order concepts. It is a commodity to be considered from a logistics and shipping point of view. It is a component for a vehicle to enable it to enjoy mobility. It is a material that is combustible and may exacerbate weapon effects on buildings that contain fuel. It is also a key component in a number of makeshift weapons employed by irregular troops (the molotov cocktail).

Tags: , , ,


Blogger chuck said...

A further thought that I believe needs to be brought out in greater highlight - Concept essentially contains the idea of not only entity and relationship components, but also the properties that define those components. In the case of entities, it is primarily the comprising characteristics that make up the different aspects of the entity, and in the case of relationships, it is that plus the specific conditionals that allow the rules component to be applied (for instance, the

28 June, 2005 21:08  

Post a Comment

<< Home