Geoscience Reference
In-Depth Information
is largely algorithmic, involving data processing, while an ES uses heuristic reasoning and infer-
ence to solve a problem. Finally, the output from a conventional system is a result, for example,
the answer to an equation or the optimal parameter found within a search space, while an ES can
generate one or more decisions with different degrees of certainty, for example, mine mineral
x
in
location
y
with a certainty of
z
.
11.3 WHEN TO USE AN ES
Determining whether an ES is appropriate for a particular problem is probably the most important
question that should be asked before following such an approach. Medsker and Liebowitz (1994)
have defined 42 guidelines for selecting an appropriate problem for ES, while a simpler framework
has been provided by Giarratano and Riley (2005). The latter authors define five critical questions
that should be asked before embarking upon an ES project. These include the following:
Q1. Can the problem be solved effectively by conventional programming?
If the answer is yes, then an ES is not the best choice since ES are more suitable for situations in
which there is no efficient algorithmic solution. Such cases are ill-structured and semi-structured
problems for which reasoning may offer the only possibility for a good solution.
Q2. Is the problem domain well defined?
It is very important to have a well-bounded problem domain so that it is clear what the system is
expected to know and what its capabilities will be.
Q3. Is the problem-solving knowledge heuristic and uncertain?
The expert knowledge may be a trial-and-error approach rather than a method based on logic and
algorithms. If the problem can be solved simply by logic and algorithms, it is better to use conven-
tional programming and not an ES.
Q4. Is there a need for an ES?
When a process is complex and time-consuming and requires some automation or when transfer-
ence to a non-expert domain is required, then there is likely to be a need for an ES.
Q5. Is there at least one human expert who is willing to cooperate and are experts able to transfer
the necessary knowledge?
There must be an expert who is willing, and preferably enthusiastic, about the project who is able to
explain the knowledge in explicit terms. Not all experts are willing to have their knowledge exam-
ined for faults and then captured in a knowledge base. Even if there are multiple experts willing to
cooperate in the development of an ES, Giarratano and Riley (2005) suggest that it would be wise
to limit the number of experts involved in the development process.
11.4 BASIC COMPONENTS OF AN ES
A typical ES consists of three main components (Figure 11.1): the user interface,
which is respon-
sible for the communication between the system and the user; the
knowledge base, which contains
the knowledge about a problem domain, and the
inference engine, which carries out the reasoning in
order to reach a solution. The user provides the system with facts, which include information regard-
ing the problem in the form of a database, and/or user preferences in the form of answers given to
the questions posed by the system. Other relevant data may be also provided to the system in the
form of databases. The inference engine (or control mechanism), which is the brain of the system,
then links the knowledge base with the supplied facts to draw conclusions. These are then sent back
as expert advice or expertise to the user via the user interface (Schnupp et al. 1989).
In addition to these basic components, an ES may also provide an explanation facility,
which is
able to justify the reasoning behind a decision; and a knowledge acquisition facility,
which offers
