Expert System is a computer program designed for modeling / simulation capabilities make an expert in solving a problem. Expert system is one of information systems classification. Two important capabilities of an expert who tried to be modeled on the expert system is knowledge (knowledge) and the concept of thinking (reasoning) of the expert. Expert system has two important capital, which is “knowledge base” and ”Inference engine” which is used to interact with the user (user interface) to be able to produce both these capabilities.
“Knowledge base” contains very specific knowledge provided by an expert to solve specific problems. For example: knowledge of a specialist to diagnose certain diseases. ”Knowledge planning” is provided by an investment consultant. ”Inference engine” is the “engine” knowledge processing, which is modeled based on the concept of knowledge to think of expert knowledge providers. Inference engine along with information obtained from a problem, paired with the knowledge that is stored in the “knowledge base, trying to find / draw conclusions, answers and recommendations to solve the problem.
Goal of building an expert system are :
1. Replace an expert
2. Possible to get expertise after working hours or at other locations
3. Automation of routine work that requires expert
4. Expert retire or die
5. Expert expensive
6. Expertise required in dangerous places (hostile environment)
7. Helping an expert
8. Routine tasks, increasing productivity
9. Completing difficult tasks, so it can be more effective in controlling complex problems.
There are two methods used to identify problems in the expert system are :
1. Forward Chaining.
A problem-solving method used to obtain the solution of a problem based on existing conditions. Existing conditions are input to the Knowledge Base Systems (KBS). KBS will process input based on information they have and produce an output which is the result of the withdrawal of the forward chaining inference. Overview of the forward chaining can be seen in the picture below.
Specification : IS : Inference Systems (System Drawing Conclusions). Forward chaining method to see the existing conditions and predict what will happen / What events caused by current conditions. Conditions now are the causes and conditions then the result or conclusion of the previous conditions. Forward chaining moves forward by looking at the cause and find a result or conclusion.
2. Backward Chaining.
A method for finding the facts in a way that there is subgoals traced recursively. The fact is that the goals and subgoals have been not always be true. The following image is a backward chaining scheme.
Backward Chaining Scheme
Specification : IS : Inference Systems (System Drawing Conclusions). Backward chaining starts with the goals sought to know the variables and then do the searching variables in all rule THEN statement. All variables in the IF condition in the rule subgoals. Subgoals value searched for in the THEN statement in any other rule, if not found in the rule, sought the direct input from the user on the ASK statement.