The ICT Lounge
 
Section 7.17:
Expert Systems
 
In this section we are focusing on expert systems. You will look at the following:
  • What expert systems are
  • What expert systems are used for
  • How expert systems are created
  • How a typical expert system would be used
  • Some examples of expert systems
Key Concepts of this section:
#
Understand what expert systems are and what they are used for.
#
Know and be able to describe the steps that are used in order to build an expert system.
#
Be able to discuss the advantages / disadvantages of using expert systems.

Expert Systems - An overview
Key Words:
Mimic, Expert, Query, Knowledge Base, Inference Engine, Rule Base, User Interface
What is an Expert System?
#
Expert systems are special databases which are designed to mimic (copy) the expertise and knowledge of a human expert in various different subjects.

Examples:
Expert systems store expert's knowledge on a range of different subjects. People can then question (query) the system to access this knowledge
(click image to zoom)
Expert systems allow doctors to make accurate diagnosis of a patient's illness
The exact species of animals can be identified
using expert systems
 
Experts are interviewed and their knowledge is gathered and put into a knowledge base.
 
An expert system rule base being programmed
(Click to zoom)

A user interface with option buttons for the
user to answer questions

Expert systems are thoroughly tested to make
sure that their rules come up with the correct
answers and decisions
 
Sometimes it is obvious why a car won't run but for less obvious problems a car diagnosis expert system can help recommend what you need to do
#
Knowledge from human experts is entered into the system and stored.

#
People using the expert system can then question it (query) and access this stored knowledge.

The system searches it's stored data for anything that provides an answer to the information that a user has input.

The expert system can provide the answers to difficult problems in the same way that human experts could!!


Examples of different uses of expert systems
#
The uses of expert systems include:
  • Diagnosing a patient's illness
  • Searching for oil and minerals reserves
  • Chess games
  • Identification of plant and animal species
  • Diagnose faults with cars
  • etc.....

More on these later where we look at some of the examples in more detail...........


How are expert systems created?
#
Expert systems are set up in 5 steps (using 3 components). These steps can be abbreviated to IKIUT:

1
Interview the experts
Experts in a field are interviewed and their knowledge is collected.

2
Knowledge base is created (component 1)
All of the knowledge collected from the experts is organised and stored in the knowledge base.

The knowledge base is searched by the inference engine (see point 3)

3
Inference Engine is created (component 2)
This allows the knowledge base to be searched by the user.

The inference engine poses questions to the user and then analyses the answers by running a set of rules that have been programmed into the engine.

Examples of questions that a medical diagnosis expert system might pose to a user:
 
  • Do you have a runny nose? YES/NO
  • Do you have a sore throat? YES/NO
Examples of rules that allow the inference engine to come up with an answer or a conclusion to a problem:
  • IF the patient has a runny nose AND a sore throat THEN the diagnosis is a cold
  • IF the animal has 4 legs AND a tail AND barks THEN the species is a dog.

4
User Interface is created (component 3)
This allows the user to communicate with the expert system. The user interface gives the user the ability to answer questions posed by the system. Common ways of doing this are:
  • Text boxes
  • Check boxes / Option buttons
  • Submit buttons

5
Testing
Finally the expert system is tested to make sure that the correct answers to problems or questions are being generated.


How would a typical expert system be used?
#
The table below shows how a car diagnosis expert system would be used:

1
A user interface screen presents user with a set of questions relating to different car problems.

NOTE:
You can try out an example of a car diagnosis expert system by clicking here

2
The user answers the questions by typing or selecting option buttons / check boxes

3
Inference engine uses the rules base to match user's answers to the data stored in the knowledge base

4
The system diagnoses the problem with the car and suggests how the user can fix it.


Advantages / Disadvantages of Expert Systems
Advantages
Disadvantages
Fewer mistakes - expert systems never forget answers to problems. People do!
They cannot make judgments and lack common sense.
More knowledge than a single human expert (combine the knowledge of many experts).
Errors in the rule base can lead to incorrect decisions being made.
Cheaper to use than hiring very expensive human experts to solve your problems.
Require lots of training before people can use them correctly.
More consistent answers than human experts (they give the same answers to the same problems every time).

Examples of Expert Systems
Expert systems can be used for a wide range of purposes. Here are a few of them:
Medical Diagnosis
Expert systems in doctors surgeries and hospitals
Medical diagnosis expert systems can help doctors to figure out what is wrong with patients. They also suggest possible treatments.
 
NOTE:
NHS Direct is an online self-diagnosis expert system. You can see how it works here

#
Expert systems are used in doctors surgeries and hospitals to help medical professionals diagnose patients correctly.

#
These expert systems make it easy for doctors to match symptoms to probable causes. they also recommend what action the doctor should take (e.g. correct medication)


Expert systems within self-diagnosis web sites
#
Some health services have web sites that allow users to self-diagnose themselves.

#
The expert system guides the user through a set of questions which are answered by typing in boxes or clicking options.

#
The system will suggest a diagnosis and the user can decide if it is serious enough to call a doctor or if the health problem can be solved with a an asprin and rest.


Locating minerals and oil reserves
#

Expert systems can be used to figure out the most likely locations to find new reserves of minerals, oil and gas.

Expert systems can help predict the best
places to drill for oil
#
The expert system's knowledge base contains geological information such as rock formations that could contain oil, minerals and gas etc.

#
The expert system would advise on the chances of finding the oil, minerals or gas at certain locations.

#
Companies can use this advice to decide on the best spots to set up rigs (e.g. oil rigs).


Car fault diagnosis
#

Modern cars are complex and when they go wrong it can be difficult for mechanics and engineers to know exactly what the problem is.

Expert systems help diagnose car faults
#
Car manufacturers create expert systems which contain lots of information about their cars in a knowledge base.

#
The expert system can guide the mechanics / engineers through different tests until the exact problem can be found.
#
The system will also recommend how to fix the car's fault.
 

Chess Games
#

Chess computer games are considered expert systems because the game mimics (copies) an expert human player.

Some computer expert system chess games are so good they have beaten human masters
NOTE:
Click here to see a video of a computer expert system beating a grand master at chess.

#
The chess game (expert system) has a knowledge base which contains different possible moves and strategies.

#
When the human player makes a move the computer would respond by checking the knowledge base for the best possible counter move.

#
When the best counter move has been found, the computer would move a chess piece into the best possible position.
   
   
 
Previous -
Next -