Expert System and Its Major Types

Knowledge-Based

It is known as the heart of an expert system. It contains the expert knowledge of a particular domain. A knowledge-based it contains both the general knowledge and case-specific information. “Knowledge representation is the process of choosing an appropriate knowledge formalism, in which the acquired knowledge is encoded so that it becomes amenable to computer manipulation.” In the case of a rule-based expert system, if-then rules are used to represent knowledge of a particular application domain. It is compared to a particular domain situation, only after that the experts reach conclusions of the problem, and also recommendations can be made. It may also include the meta-knowledge of the domain.

Inference Engine

It is an interpreter for the knowledge base. And it provides the knowledge to the solution of a particular problem. The knowledge from the knowledge base is used for solving. New knowledge is been derived as it is not explicitly stored. It can answer questions from users; additional information is requested from the users if it is necessary for answering the question.

Explanation Generator

It provides the reasoning for a particular program, as an expert system should be able to explain the reasons for a specific problem. The explanation contains the justification for the problems’ conclusion, just like ‘why the system needs a particular piece of data, ‘where useful’, ‘tutorial explanations’ etc. If convinced reasoning and accuracy are not present people will reject the advice. For a good explanation arguably there are three aspects as epistemology, user modeling, and rhetoric, which are tricky for computer systems. Criticisms are there for expert systems for their poor explanations.

User Interface

Are used for efficient user interactions with the system, which simplifies the communications and it reduces the complexity. The interface may be of question-and-answer type, menu-driven type, or graphical interface style, etc. The decision is finally taken only when the user needs and requirements of the knowledge-base and the inferencing system compromises.