Expert Systems
What is an Expert System?
An expert system (also known as a knowledge-based system) is able to store and manipulate knowledge to help the user solve a problem or make a decision.
Typically, the system runs by understanding and applying rules. For example, a careers service might use a system like KUDOS which asks people questions about their likes and dislikes, strengths and weaknesses and it can then suggest some careers that the person might be interested in pursuing an interest in. Systems have been developed for medical diagnosis where symptoms can be entered and an illness is diagnosed.
http://www.caughtspeeding.co.uk/ contains an expert system that asks you questions about your speeding conviction and then tells you how the court is likely to deal with you.
Expert systems are also used by social services (to calculate benefit payments) and by financial services (who want to predict market movements)
An expert system allows expertise that is in short supply to be widely shared.
See also the example of ELSIE (Heathcote p.228) and the Nemesis system (p.230-1)
Typical Features of an Expert System
It can reason with "uncertain data" e.g. a "don't know" response from the user.
It delivers advice to the user.
It explains its reasoning to the user.
Expert systems are best suited to situations where a well defined set of rules exists.
Expert System Shells
This is a program that allows someone to develop an expert system without knowing a programming language. The "expert" inputs his knowledge into the system. For example, a medical team could input lots of symptoms into a system that can then apply rules to diagnose illnesses.
Benefits of Expert Systems
In businesses, expert systems can perform some tasks much faster than a human being e.g. calculating a benefit claim
Expert systems can quickly identify faults in equipment
Successful expert systems are very accurate - maybe more accurate than a human being and probably cheaper than using the time of a highly paid professional.
Recommendations will be consistent and always impartial.
An expert system can be used to harness the combined knowledge of lots of experts.
Limitations of Expert Systems
Expert systems do make mistakes and they do not learn from them.
Expert systems work best when the problem is specific and well-defined. They like making structured decisions where clear rules can be followed. They are less suited to less predictable decisions.
Human beings could follow the expert system slavishly. It is crucially important that people interpret the results of an expert system.
A human advisor may take into account special circumstances which an expert system may overlook.
Although the expert system should give its reasons for coming to a particular decision, it cannot be questioned any further.
Case Study
See the example of Scottish judges (Mott and Leeming p.144)