2.1 Concepts of Knowledge and Knowledge Representation
2.1.1 The concept of knowledge
Knowledge reflects the relationship between things in the objective world , and different things or different relationships between the same things form different knowledge .
Knowledge in AI falls into two categories: facts and rules . Fact : " Snow is white ". Reflects the relationship between "snow" and "white". The rule : " If you have a headache and runny nose, you probably have a cold ." reflects a causal relationship between "headache and runny nose" and "you probably have a cold."
2.1.2 Characteristics of knowledge
1. Relative correctness
Knowledge is generally correct under certain conditions and circumstances .
2. Uncertainty
1) Uncertainty caused by randomness
2) Uncertainty caused by ambiguity
3) Uncertainty caused by experience
4) Uncertainty due to incompleteness
3. Representability and availability
The representability of knowledge means that knowledge can be expressed in an appropriate form, such as text, language, graphics, neural network, etc.
2.1.3 Classification of knowledge
1. According to the scope of knowledge , it is divided into common sense knowledge and domain knowledge
2. According to the function and expression of knowledge , it is divided into factual knowledge , procedural knowledge and control knowledge
Factual knowledge is used to describe related concepts, facts, attributes and states of things in the field. "Sugar is sweet", "Xi'an is an ancient city", "a year has four seasons of spring, summer, autumn and winter". Factual knowledge generally adopts the form of direct expression , such as expressed by predicate formula.
Process knowledge mainly refers to the knowledge about system state change, operation, calculation and action of problem solving process. Process knowledge is generally regular knowledge obtained through comparison and analysis of various problems in the field , and is composed of rules , laws , theorems and experience in the field .
Controlling knowledge , also known as deep knowledge or meta-knowledge , is knowledge about how to use existing knowledge to solve problems, so it is also called " knowledge of knowledge ".
3. According to the structure and expression of knowledge , it can be divided into logical knowledge and visual knowledge
Logical knowledge: Knowledge that reflects the logical thinking process of human beings, such as human empirical knowledge, generally has the characteristics of causality and difficult to describe accurately.
Visual knowledge: Knowledge established through the image of things is called visual knowledge. For example, it is difficult to describe "what is a tree" in words, but it is easy to understand by pointing to a tree and saying that it is a tree.
4. According to the certainty of knowledge , it can be divided into certain knowledge and uncertain knowledge
Deterministic knowledge means that the truth value can be judged, and it is between "true" and "false".
Uncertain knowledge refers to knowledge that is imprecise, incomplete and fuzzy.
2.1.4 Representation of knowledge
Knowledge representation is to formalize or model human knowledge . In fact, it is a description of knowledge, or a set of conventions, a data structure acceptable to computers for describing knowledge. When choosing a knowledge representation method, the following aspects should be considered:
1. Fully express domain knowledge
2. Facilitate the use of knowledge
3. Facilitate the organization, maintenance and management of knowledge
4. Easy to understand and realize