Expert Systems

Describing an expert system:

Expert systems are transcribed human expertise into if-then rules. These rules are configured to better serve a novice who may not have the capabilities to have contact with a professional due to physical proximity or any limiting factor. It means that someone who has basic knowledge of the computer to simply input certain variables that the system requires can be provided an output of information that has an illustrated record of success in practice. Below is the father of the expert system Edward Feigenbaum. He talks about several systems he has helped to develop that provide assistance to professionals in the healthcare industry.

The parts of an expert system (technical components):

An expert system consists of 5 components:

-The user interface; The user interface is just an overlay that accesses the functions of the systems, and should be well displayed and easy to interact with.

-The working memory; The working memory represents the set of facts known about the domain.

-the knowledge base; The knowledge base analyzes the domain to alter the working memory to satisfy conditional arguments, usually in an if-then format with a number of antecedent arguments and consequent actions that range from altering the working memory to displaying output to the user.

-the inference engine;  Then there is the inference engine which is the program part of an expert system, it is a problem solving model that uses the rules of the knowledge base and the case-specific knowledge of the working memory to solve a problem. A rule is "fired" when all the antecedents for that rule are met. Backward chaining reasoning means the system has reasoned backward from the goal to be derived.There is forward chaining that goes on until all antecedents are true for a given cycle.

-and the explanation system; The explanations system allows the user to see the reasoning of the system and 

the rules (knowledge) used.

Expert Systems in Business:

DENDRAL- called the first expert system, identifies molecular structure of unknown compounds.

MYCIN- Made significant contributions to the field, not used in practice but helped to diagnose and design treatments for meningitis and bacterial infections

PROSPECTOR- Used to locate deposits of ore like copper and uranium  Actually used in SRI international rather than Stanford, as the previous systems. 

ALTREX- helps diagnose engine troubles in Toyota cars.Can be called up by service technicians in the plant.

PREDICATE- An expert system that helps to analyze and provide estimates of the required amount of time to construct high rise buildings.

Here is a valuable resource.

Benefits/Challenges

 What's unique about an expert system is that it has the capability to explain it's reasoning, also it has to do with a limited domain of knowledge. It remains relevant to a specific purpose but not general enough to replace an actual expert in a field in the range of domains that a human system can develop. The ability to comb through massive amounts of data points to 'understand' conditional triggers and extrapolate them into a solution unique for that case also, most importantly, while providing reasoning to every logical step along the way.