5.3.3 Intelligent Knowledge Based Systems (IKBS) and expert systems

The purpose of an expert system / Intelligent Knowledge-Based System is to use 'artificial intelligence' to replace a human 'expert' in a particular area.  This requires a system that can process information into knowledge.

For a system to have 'knowledge' means it must show some level of understanding of the data that it stores.  For this to happen, the system must be programmed  with 'rules' so it can give some meaning to the information it finds.

To create such a system would require the following:

• a large database of data on the specific area (i.e. data on medical symptoms and diseases).
• a method of searching the database based on questions the system asks, and the responses the users gives (i.e. by taking a patient through a series of questions on their symptoms).
• a set of rules for making deductions from the responses given and the data in the database (i.e. the system makes a diagnosis based on the symptoms).

Examples:

1. A medical diagnosis expert system could be used in a doctor's waiting room. Patients would use a touch screen to answer questions on symptoms etc. created by the system.  Based on the patient responses, the system could use it's database of diseases and symptoms, along with it's programmed rules, to prepare a list of possible diagnosis for the doctor to investigate further.

Advantages.

• The doctor saves time because they do not have to ask the patient to describe their symptoms in person.
• The doctor is given a suggested list of possible diagnosis to investigate further.
• The computer can store far more information than the doctor and can search it far faster and more efficiently.
• The database can easily be updated or extended.

Disadvantages

• It can be difficult to describe symptoms to a computer system.
• It relies on a basic level of skills from the user.
• It lacks the 'human touch' of a doctor actually talking to a patient.

2. A trouble-shooting program could help diagnose what is causing problems with software or hardware.  The system would ask a series of questions and give instructions on possible actions to try out.  If the actions were successful the program would end but if not then it would ask further questions and suggest further actions.

Advantages.

• The user saves time because they are taken through a logical series of things to try out that are likely to solve common problems.
• The system can directly access areas of the software such as printer or network management screens without having to give complicated instructions.
• The computer can store details of a huge range of common computer problems.
• The database can easily be updated or extended as new problems are identified.

Disadvantages

• It can be difficult to answer questions on something the user may know nothing about.
• It relies on a basic level of skills from the user.
• It lacks the 'human touch' of being able to discuss the problem in everyday language.

3. A expert system used by a car mechanic could help to diagnose faults in a car by asking the mechanic to carry out tests or answer questions.  This could also be automated as the computer could have inputs from sensors or have a direct interface with a computer system built into the car. The system would then give the mechanic a list of probable  faults with the car even make automatic adjustments using the computer system built into the car.  It would also be helpful to a newly qualified mechanic as they would have access to a wider range of knowledge, could save time compared to having to contact an expert and the system could be used as a training aid.

Advantages.

• The user saves time because the mechanic is taken through a logical series of things to try out that are likely to solve common problems.
• The system can directly access computer systems built into many modern cars.
• The system can store details of a huge range of common faults with different makes of cars.
• The database can easily be updated or extended as new problems are identified.

Disadvantages

• It can be difficult to answer questions on something the user may know nothing about.
• It relies on a basic level of skills from the user.