A rule-based expert system is a system that is characterized by a knowledge base with the domain knowledge fed or represented in terms of rules. A rule-based system is ideally a representation of the deep knowledge of human beings. It is very complex and is comprised of different explanation facilities, which aid its functioning. The following are some of the facilities that are found in a rule-based expert system.
The first explanation facility found in a rule-based expert system is the user interface. This mechanism is used in supporting exchange of information between the system and human beings. It may be represented as a complex high-resolution display or it may just be a simplified display with a text-orientation. User interface is determined during the development stages of the system. In recent days, user interfaces of a graphical nature have been most preferred since they are user-friendly.
The second explanation facility of a rule-based expert system is the explanation facility. This is a facility whose purpose is to give a detailed explanation to the users on how the system reasons. The facility tracks the fired rules and offers interconnected reasoning that makes it possible for particular conclusions to be made. As a result, the facility is also known as the justifier. It significantly distinguishes conventional systems from rule-based expert systems. Most commercial rule-based expert systems give explanations on data fed into the machine. In some cases, the systems give an explanation of the knowledge base only while others include an explanation of the control strategy (Biondo, 1990).
The third explanation facility of a rule-based expert system is the inference engine. This is a facility whose work is to make inferences just as the name suggests. The inference engine makes decisions on rules that have satisfactory facts, categorizes them depending on importance and then works on the rules whose priority is high. Inferences can be classified into backward chaining and forward chaining. Forward chaining starts its process of reasoning by looking at the facts then moving to the conclusion while backward chaining starts with a hypothesis and then looks at the supportive facts.
The design of the system is the most important determiner of whether an inference engine entirely relies on either backward or forward chaining. The design itself is determined by the nature of problem being solved. CLIPS and OPS5 are some of the popular systems that perform forward chaining. EMYCIN on the other hand is a typical example of a backward chaining system. KEE and ART are unique systems that are known to offer both backward and forward chaining. The most appropriate application of forward chaining is control and monitoring, and prognosis while backward chaining fits in general diagnosis of problems. Inference engine performs its work in turns, performing combined tasks till the execution is stopped by particular criteria.
The fourth facility of a rule-based expert system is a knowledge acquisition facility. This facility gives the users a chance of feeding the system with knowledge thus eliminating the need for knowledge engineers to enter the knowledge. This feature is not a mandatory one in most of the rule-based expert systems. The knowledge acquisition facility is also referred to as production memory since if-then rules are referred to as productions. Through this facility, it is possible to employ rule induction method to create ordinary rules.
Technology has been evolving and in the recent days, technological advancement has been registered. Rule-based expert systems have made it possible for many procedures to be performed, procedures that would not have been performed with the traditional systems. Rule-based expert systems use rules, which make it possible for the systems to generate a chain of reasoning that eventually leads to a conclusion. Explanation facilities are not adequate on current systems since applying the rule-based system concept is not an easy thing. A technology requires a lot of investment and knowledgeable individuals in order for it to succeed. For instance, if a single mistake is committed during the design of the systems this could lead to serious problems.
Another reason why explanation facilities in rule-based expert systems are not adequate is due to the costly nature of the materials required to develop the systems. The cost is so high such that most companies opt to use conventional systems instead of rule-based expert systems. In addition, there some tasks that are well performed by the conventional systems hence there is no need of replacing them with rule-based expert systems. However, due to the changing nature of technology most organizations are moving towards the direction of rule-based expert systems.
Reference
Biondo, S. (1990). Fundamentals of expert systems technology:Principles and Concepts. New York: Intellect Books.