Expert systems were among the most exciting computer applications to emerge during the 1980s. The ES technology derives from the research discipline of Artificial Intelligence, a branch of computer science concerned with the design and implementation of programs which are capable of emulating human cognitive skills such as problem-solving, visual perception and language understanding. An expert system mimics human expertise in a narrow domain to solve specific problems in a well-defined area.
Below are the diagram of expert system with its component
An expert system has been defined differently by different people. As such, there is no precise definition of an expert system that is guaranteed to satisfy everyone. However, while in a narrower perspective, ES technologies make computer programming easier and more effective, in a broader perspective, ES represents the first step in a process that will transform computing by moving programming technologies beyond mathematical programming into a realm of logical, symbolic programming. More specifically, the following definitions could be noted. According to Peter Jackson, “An expert system is a computer programme that represents and reasons with knowledge of some subject specialist to solving problems or giving advice.”
Robert Bowerman and David Glover have defined the expert system as “highly specialized computer systems capable of simulating that element of a human specialist’s knowledge and reasoning that can be formulated into knowledge chunks, characterized by a set of facts and heuristic rules.” (Heuristic rules are rules of thumb accumulated by a human expert through intensive problem-solving in the domain of a particular task).
According to Bruce Buchanan and Reid Smith, an expert system is a computer program that —
i. Reasons for the domain-specific knowledge that is symbolic as well as numerical.
ii. Uses domain-specific methods that are heuristic (plausible) as well as following procedures that are algorithmic (certain).
iii. Performs well in the problem area.
iv. Explains or makes understandable both what it knows and the reasons for its answers.
v. Retains “flexibility.” An expert system, according to Hossein Bidgoli, “is a series of computer programmes that attempt to mimic human thought, behavior in a specific area that has successfully been solved by human experts.”
To be effective and improve the quality of problem-solving, an expert system should possess the following capabilities:
i. Capturing of expertise.
ii. Codifying the expertise.
iii. Duplicating and transferring the expertise.
iv. Saving the human expert’s time.
v. Saving on maintenance and updating of the knowledge base.
Ralph Stair and George Reynolds have identified the following characteristics of the expert system:
1. Ability to explain their reasoning or suggested decisions.
2. Ability to display “intelligent” behavior.
3. Ability to conclude complex relationship.
4. Ability to provide “portable knowledge.”
5. Ability to deal with certainty.
6. Not widely used or tested, due to the difficulty of use.
7. Limited to relatively narrow problems.
8. Inability to deal with “mixed knowledge.”
9. Inability to refine own knowledge base.
10. Difficult to maintain.
Expert systems can be used to solve problems in practically every field and discipline. Such systems can also help in various stages of the problem-solving process. As such, expert methods have been developed for a variety of complex applications. A few illustrative applications of expert systems are
1. Aerospace technology (NASA)
2. Airline/civil aviation (scheduling/routing)
3. Banking and finance (credit card limits, etc.)
6. Food industry
7. Health-care management (e.g., diagnosing blood infections) 8. Manufacturing design and assembly
9. Geological data analysis and interpretation for oil exploration drilling site
10. Personnel management
11. Security analysis/portfolio management
12. Tax planning
13. Foreign exchange management
14. Gene-cloning experiments
15. Troubleshooting telephone network
16. Configuring computer systems
17. Strategic goal was setting
18. Quality control and monitoring
1. Expert systems function in the domain of extracted, cognitive, logical thinking process. As such, ESs are not adept at managing highly sophisticated sensory inputs.
2. As ESs are based on a narrow range of codified domain, they may not be able to tackle multi-dimensional problems.
3. Due to the narrow range of knowledge incorporated in the ESs, they typically do not respond well to situations outside their range of expertise. Hence, they remain what they are Machine Experts.
4. Atypical ES may not be able to make available common sense knowledge and broad-ranging contextual information/s.
5. ESs typically lack human self-awareness and self-analysis tools. Introspection is not available as ESs also happen to be “non-self-referral” systems.
6. If a problem is not specific and has not been solved previously by an expert or some experts, then that question is not considered suitable for the expert systems’ implementation. ESs are capable of performing only within a specific, logical-oriented realm of expertise, and herein lies the major limitation of the expert systems, as computers only have memory and not necessarily, intelligence.
When to go in for expert systems?
Expertise, as we all know, has both its price as well as value. No wonder then that it is difficult, expensive and time-consuming to develop sophisticated expert systems. It would, therefore, be desirable to weigh the following aspects/ considerations before an organization decides to opt for an expert system:
1. Will the system help reduce risk significantly?
2. Will the system provide a high payoff?
3. Will the system performance be more consistent than human experts?
4. Will the system enable the expertise to be made available at multiple locations simultaneously?
5. Is the knowledge rare or expensive?
6. Will the system allow developing the solution faster than human experts?