First 708 words of the document:
N.B: Questions on expert systems are on the second exam paper but other systems questions (e.g., on systems
lifecycle) are on BOTH exam papers.
What Is An Expert System?
An expert system is a program that mimics the intelligence of a human expert in one specific field of knowledge, e.g.,
an expert system could be set up to give medical advice. The computer is given a base of knowledge from the
expert and then given the rules for processing the information (If the person says this, do this, if the person does
this, do that) The idea is that the computer asks questions similar to those that the expert would ask and then comes
to a conclusion that is similar to the conclusion the expert would come to. It can be used by a non-expert in a
particular subject area to help them evaluate and make a decision. Expert systems are useful tools, but they are
nowhere near as versatile as a person.
An expert system is a program which aims to bring together human expertise in one knowledge base (set of data
and rules). Expert systems are sometimes called "knowledge based systems" or "information based systems" (IKBS).
The solution is given by the computer asking question and the person voicing the questions and typing in the answers.
It is similar, not like, "20 Questions".
Typical Examples Of Expert Systems Are:
Medical diagnosis: To work out what is wrong with a patient
Car engine fault diagnosis: To work out what is wrong with a car
An expert system has two parts:
A Knowledge Base: Facts and rules provided by the experts
An Inference Engine: The computer program that works out the diagnosis or advice by checking the rules,
knowledge base and the input from the user. This is the part that is similar to 20 Questions.
The main features of an expert system are:
Only one specialised area (for example, medical diagnosis) is covered by the expert system.
There are many rules specified. For example, if the patient has spots and a temperature then measles is a
probable diagnosis (for example, the strength of belief that it is measles is 50%). Many rules similar to this
with different probabilities (strength of belief) could be specified.
The user is often asked to respond to questions output by the system, which cause certain rules to be
triggered. The user can provide "don't know" responses and give the degree of uncertainty attached to the
Advice and diagnosis may be given by the system.
Explanations may be provided by the system to support other output.
Reasoning is part of the processing to be carried out by the system.
The Steps For Creating An Expert System:
(this is similar to creating a system)
1) Interview experts and use other expert sources such as textbooks to gather as many facts and rules as
possible (we interviewed the designer at Benton Dore)
2) Design the knowledge base (we created the designs for the forms, reports, mail merge and switchboard)
3) Select software to use. This may be an expert system shell (an already built inference engine) or a computer
language appropriate for building the knowledge base and an inference engine (we did this for our project
2. Decided on software and hardware)
4) Implement (make) the system making sure the inference is easy to use (we made this as easy as possible to
use for the secretary)
5) Test the system (we tested the system)
6) Document the system and create a user manual (we created a user guide)
7) Check the system with the experts to make sure it produces sensible advice or diagnosis (we collected
feedback from people using the system in questionnaires)
Creating an expert system is very time consuming and the fact finding stage is often very difficult because the
systems consultant (the computer programmer making the system) often finds it difficult to understand the area of
Other pages in this set
Here's a taster:
E.g. a computer programmer making an expert system to diagnose medical
problems will find it difficult to understand all the expert knowledge he is trying to include in the expert system which
makes creating the expert system difficult.
Here are some related past paper questions to try:
1) A car engine has a fault. The car mechanic uses an expert system to try to find the solution to the fault.…read more