How Expert Systems Work
Expert systems are generally used to peform a task related to some
goal-oriented or problem-solving activity. Typical tasks are diagnosis,
planning, scheduling, configuration and design. The field of expertise
captured in an expert system is called the task domain.
The power of expert systems depends upon the specific, high-quality
knowledge it contains about task domains. Current knowledge acquisition
methods are slow and tedious, hence much of the future of expert
systems depends on improving knowledge acquisition techniques and
in codifying and representing a large knowledge infrastructure.
Creating An Expert System
Expert systems are created using processes known as knowledge engineering
by knowledge engineers. The knowledge engineer must:
- make sure the computer has all the knowledge needed to solve
- choose one or more forms in which to represent the required
knowledge as symbol patterns in the memory of the computer - a
- ensure that the computer can use the knowledge efficiently by
selecting from a handful of reasoning methods.
How An Expert System Works
An expert system works by capturing the intelligence of an expert
[within the task domain], capturing the expert logic and rules in
a software program, and creating a user interface to access this
logic. The process would:
- Real experts in a defined problem domain are asked to provide
"rules" on how they normally evaluate problems in their
- Scenarios are then created based upon different combinations
- Individual attributes are identified, and any dependencies or
other relationships are noted.
- Systems developers will then enter these logical entities, and
link the relationships.
- Test cases are built and evaluated by the experts. Acomputer
program can then be used to examine this test data to derive rules.
- The end-user interfaces with an expert system through an interactive
Simple expert systems use simple true/false logic to evaluate data,
but more sophisticated systems are capable of performing at least
some evaluation taking into account real-world uncertainties, using
such methods as fuzzy logic. This type of sophisticated intelligence
is difficult to develop and still highly imperfect.
Most expert systems are self learning, based on the pattern of
inputs and responses by users. The logic becomes more and more valid
with each cycle of use.
Using An Expert System
Expert system dialogs are not pre-planned. There is no fixed control
structure. Dialogs are synthesized from the current information
and the contents of the knowledge base. Supplying an the answer
to a particular questions does not stop the consultation process.
Different types of user answers prompt different types of responses
from the expert system. For example - if the user answers "Why"?
The expert system may attempt o explain their path of logic:
I am trying to determine the type of 'x' to suggest. So far 'y'
is not a likely choice. It is possible that 'z' is a likely choice.
I know that [variable] and the preferred [requirement] is [attribute]
then there is strong evidence that the [objective] choice should
Traditional systems do not accomodate answering questions such
as 'how' or 'why'. Expert systems will respond by stringing together
the the underlying knowledge structure. This structure is linked
by program rules, values and known associations.
A rule is set of antecedent conditions which, if true, allow the
assertion of a consequent.
The rule references values, and tests them against various constraints
or asserts constraints onto them.
Values may be associated with some organizing entity. For example,
an individual traveler is an entity with various attributes (values)
including whether they prefer to fly or travel by train, and what
class of seat they prefer.
There are also rules, which associate the currently known values
of some attributes with assertions that can be made about other
attributes. It is the orderly processing of these rules that dictates
the dialog itself.
NEXT: Expert System
Back To Top