5. Knowledge Representation-Object-Oriented Knowledge Representation

The basic starting point of the Object-Oriented Knowledge Representation Method is that the objective world is composed of some entities. These entities have their own States and can perform certain actions. Similar entities are abstracted as high-level entities, which can be associated in some way. Objects are images of these entities. The object encapsulates data members (or instance members) and member functions (methods ). Data members can be used to describe various attributes of an object, which are hidden from the outside. External users can access private members of an object only through member functions. Data members can be initialized and changed through member functions. Therefore, the object can dynamically save its current state. Because the object also contains operations (member functions), the solution mechanism can be encapsulated in the object. In this way, an object is both a storage unit of information and an independent unit of information processing. It has a certain internal structure and processing capability. Various types of solving mechanisms are distributed across various objects, and the entire problem solving process is completed by passing messages between objects. The knowledge represented by objects is closer to the objective situation. This representation scheme is more natural and easy to understand.

The object-oriented representation method has the following advantages:

• inheritance brings about natural hierarchy and structure. At a high level, objects can encapsulate complex behaviors to make the specific details transparent to the knowledge usage at the layer, thus reducing the complexity of Problem description and computational reasoning; inheritance can reduce the redundancy of knowledge expression. It is convenient to modify, add, delete, use, and maintain a knowledge base. modifying a Knowledge Unit does not affect other units, the knowledge rules contained in each Knowledge Unit are limited, and the reasoning space is small, which improves the reasoning efficiency.
• the definition of the object produces good compatibility and flexibility. It can be data or method; it can be a fact, a process, a framework, or a semantic subnetwork.
• in geometric language, the object-oriented abstraction mechanism is actually to regard the object as the fragtal element of the objective world and its ing system, so things can all be formed by these analysis elements. Classification features are constantly subdivided, which is consistent with the continuous expansion of the knowledge structure. The second is "proportional self-similarity", which makes it possible to "derive complicated systems from simple principles ".