Information Technology Reference
In-Depth Information
Fig. 11.1
The interaction
between the different types of
inferences
Retroduction
Induction
Deduction
numbers. The abstraction element for inference is thus already done. The mechanism
described here is simplified by excluding any direct and dynamic modification of
one inference type by another. The processes of inference will remain fixed for this
implementation.
The black solid circle in Fig.
11.1
is a controller that manages the flow of infor-
mation (thick arrow) between the three inferential mechanisms. A tension is created
when the three inferential mechanisms cooperate to formulate a viable hypothesis.
Communication between the mechanisms involves a cycle of retroduction, deduc-
tion and induction as well as the feeding-back of information, which will ensure that
a hypothesis is applicable within the three mechanisms. In a normal mixed IQ test
that involves other sequences than numbers, the process of 'abstraction' becomes an
important component of this tension (c.f. Chap. 2).
For a given set of facts (e.g. a sequence of numbers), it is the role of
retroduction
to create a reasonable hypothesis for those facts (e.g. that particular sequence).
De-
duction
exercises the hypothesis and returns a prediction of a new fact (e.g. the next
number in the series). The validation of the retroduced hypothesis is the function of
induction
that ensures that the hypothesis is suitable for the purpose (e.g. the predic-
tion is correct, the calculation was not too complex, it fits all or most of the facts and
the form of the hypothesis is simple).
The process of creating and validating a hypothesis is governed by:
•
a set of hypothesis generators called concepts,
•
a set of criteria, and
•
two stages of learning that orders the formation of these concepts and criteria.
The learning process and the final selection of a hypothesis from the set are dependent
upon the validation of the hypothesis against a criterion. The criterion is selected from
a set of criteria (see Table
11.1
) each of which relate to a hypothesis or concept. The
validation process (i.e. induction) is the test for specified features of the generated
hypothesis against a pre-set criterion (see Table
11.1
).
11.3
Learning in Retroduction and Induction
However, the process of inferring a hypothesis depends not only upon the cycle of
retroduction, deduction and induction but also upon two stages of learning. There is a
primitive learning scheme in both the retroductive and the inductive module. Learning
in retroduction is a process of using the features of the abstraction to choose a concept
