I.2.2.
General Concept of the expert systems
The knowledge of an expert system can be represented several
in manners (it can be encapsulated in the rules and the objects).
A common method to represent knowledge is in form of
the rules of the type IF... then...
For example:
IF the light is red THEN you stop.
A traditional expert system embarks a not written knowledge
which must be obtained from a specialist through intense interviews by
an engineer in knowledge (knowledge engineer)
for a long period. The process of the construction of an expert system is
called engineering of knowledge and consists of the
acquisition of knowledge near a human specialist or another source, and its
coding in the expert system.
General stages of the development of an expert system are
developed in the figure 1.3 here after:
Human specialist
Engineer of knowledge
Basic knowledge of the expert system
Dialogue
Clarification in rules
The engineer of knowledge establishes initially a dialogue
with a specialist to obtain his knowledge. Then, the engineer of knowledge
explicitly codifies knowledge in the basic knowledge in form of the rules. The
specialist then evaluates the expert system and present criticisms at the
engineer of knowledge. This process is repeated until the expert considers
satisfying the behavior of the expert system.
The expert systems are often conceived differently from the
conventional programs because the problems generally do not have an algorithmic
solution and depend on the inferences to obtain a reasonable solution, by
considering the solution obtained as best which one can expect if there is no
algorithm leading the optimal solution. Because the expert system depends on
the inference, it must be able to explain its reasoning so that it is checked.
The facility of explanation forms the
integral part of the sophisticated expert systems.
Some expert systems allow that the system learns from the
rules through the example by using the induction of the rules
starting from the tables of the data.
I.2.3.
Characteristics of an expert system
An expert system can be conceived so that it has the following
characteristic:
o High output: The system must have the
capacity to answer to a qualification level equal or higher than a specialist
in the field. That means that the quality of council given by a system must be
very high.
o Adequate response time: The system must
act as a reasonable time, comparable or better at the time required by a
specialist, to make a decision.
o Reliability: the expert system must be
reliable and does not have to know "faults" if not it will not be used.
o Comprehensible: the system must be able to
explain the stages of its reasoning while they are carried out, instead of
being only one block box which produces a miraculous answer.
o Flexibility: Considering the great quantity
of knowledge which an expert system can have, it is significant to have an
efficient mechanism to add, modify, and eliminate knowledge.
A reason of the popularity of the expert systems based on the
rules is the efficient and modular capacity of storage of the rules.
| 
