What is knowledge representation?
In general, knowledge representation is a set of conventions used to describe the world. It is a symbolic, formal, or modeled knowledge.
From the perspective of computer science, knowledge representation is a general method to study the feasibility and effectiveness of Computer Representation of knowledge, and a strategy to express human knowledge as the data structure and system control structure of machine processing.
A Knowledge Representation Method usually has the following basic requirements:
- Have sufficient presentation capabilities. For a specific domain, the ability to correctly and effectively express the knowledge required to solve a problem is the ability to express knowledge. This is a key issue. The selected representation method must expand the representation range as much as possible and improve the efficiency as much as possible. At the same time, information in the natural world is inherently vague and uncertain. Therefore, the degree of support for the ambiguity and uncertainty of knowledge is also an important factor to be considered during selection.
- Matches the reasoning method. Artificial intelligence can only process the knowledge representation suitable for reasoning. Therefore, the selected knowledge representation must be suitable for reasoning to solve the problem.
- The consistency of knowledge and meta-knowledge. Knowledge and meta-knowledge are different levels of knowledge. Using a unified representation method can simplify Knowledge Processing. If we know the premise, we need to export the necessary conclusions as quickly as possible and solve the problem of how to export the best conclusions. Then we can add some control information to the meta-knowledge, that is, the heuristic information.
- Clear and natural module structure: Because the knowledge base must be constantly expanded and improved, the representation mode with a disruptive structure is conducive to the acquisition of new knowledge and the maintenance, expansion and improvement of the knowledge base; indicates whether the mode is simple and effective, facilitating reasoning of domain problem solving strategies and searching knowledge bases, which involves knowledge usage efficiency; the representation method should also have well-defined semantics and ensure the correctness of reasoning.
- Descriptive representation and procedural representation: It is generally considered that descriptive knowledge involves less details and a high degree of abstraction. Therefore, it is natural to express it, has good reliability, and is easy to modify, but has low execution efficiency; the characteristics of procedural knowledge representation are the opposite.
In fact, the process of selecting a Knowledge Representation Method is a compromise between clear expression and efficient use.
Currently, many knowledge representation methods are used:
Predicate logical notation
Generative notation
Framework notation
Semantic Logic Representation
Object-oriented notation
Ontology-based Knowledge Representation
Petri Net-based representation
Fuzzy logic-based representation
Relational mode-based representation
Event-based association Representation
Decision Table-based representation
Problem reduction-based representation
