Expert systems

Expert systems are the first step to a complete artificial intelligence. But are they already used in todays and in which areas will they be used? And the most important question: What exactly is an artificial intelligence?

 

Definition

An expert system is a computer program that is able to support human like an expert in difficult questions. To do this they use a large data base around this theme. So an expert system is able to
produce and improve the data base, to solve the problem with it and to explain it to the human.
Sounds great, doesn't it?

Variants

The expert systems are defined in three main categories.

Case-based reasoning

Such an expert system builds its knowledge base by defining cases for every situation.
1. At first the system tests if the case and its description match to an already existing one in the knowledge base.
2. Then the approach gets called if the case and its solution already exist. But if the case doesn't already exist it shows you an error. Some systems solved this by calling you similar cases.
3 If you are able to solve the problem with the solution the system tells you everybody is happy. If not you have to solve it on your own and type your solution into the program.
4 The program will then improve its data base by adding your solution.

The problem of this variant is that you have to define your case very precisely to get a precise answer. And so you can imagine how big the data base gets if you work in an area as big as the medicine. That is one reason why these system work only in very specified areas and they do their job very good.

Rule based systems

A rule based system needs two types of knowledge bases, one where the facts are stored in and a rule base. A rule base contains the rules the system has to follow and to respect but it also needs them to connect the facts.
1. The system gets a problem.
2. An algorithm selects necessary rules for the problem or tries to prove a rule.
The forward chaining algorithm selects the rules and wants to solve the problem.
The backward chaining algorithm tries to prove a rule by getting facts and an already existing solution.
3. The system explains the human its solution and improves itself and its data base.

Decision tree

The latest idea is to use decision trees they are able to learn on its own. The system learns on its own by using an example volume. This system is much more complex than the others but the main goal is crate a precise tree based on attributes of the examples. The system runs step by step it's created tree from the top to the bottom and every knot of the tree symbolises an attribute for example of a patient. And in the system ends in a solution, the leaf of the tree.

And if something doesn't work new attributes can easily be added by the system.

Prospects

Expert systems can be used in nearly every area from the hospital to the flood forecasting. And this is one point which makes the development of artificial intelligence to such an important science in the future, together with the nanotechnology and some others. These technologies can be used everywhere.