Knowledge Representation

Abstract Solutions to many problems in AI depend more on the availability of a large amount of knowledge than on sophisticated algorithms. Knowledge representation is the study of ways of supplying programs with such knowledge. The term “knowledge base” is used to refer to the body of knowledge made available to the program. The two main parts of any AI system are a knowledge base and an inferencing system. The specification of a knowledge representation consists of two major components: A description of the notation used to express facts and a description of the operations that can be performed on a knowledge base. A notation in which facts can be expressed is essential to any knowledge representation. To build the knowledge base, a variety of knowledge representation schemes are used including logic, lists, semantic networks, frames, scripts, and production rules. Although there is little agreement as to what knowledge representation actually is, many schemes have been proposed as general frameworks for representing and storing knowledge. Some have been successfully used as a basis for working systems. There are, however, many features of knowledge such as defaults that are not well understood. Until a better understanding of these features is reached, knowledge representation will remain an active area of study.