Information Technology Reference
In-Depth Information
necessary to consider every option may apply in large, strong units (e.g. business
ones).
Cognitive decision systems also follow the rule of the materiality of the
adopted criterion. It is well known that not all decision-making criteria can be as-
signed the same weight in every case, as some of them are more material than oth-
ers, even when this does may not seem desirable. In addition, some decision-
making criteria may operate interactively, so they cannot be considered in total
isolation [75].
Cognitive decision systems analyse and then execute decision-making proc-
esses using the classical decision theory, so they try to eliminate new approaches
to the subject of decision-taking. These latter approaches are based on the current
decision-making strategies associated with humans. The most important of them is
the satisficing strategy which consists in reviewing the available options in a ran-
dom order and selecting the first which suffices to satisfy the decision-maker
[128], [129]. Human decision-making processes are not the best, as people taking
decisions do not compare all the available options in every possible regard to se-
lect the best one. For a human, this is very often impracticable because of the
shortage of time, knowledge, cognitive resources or computing capacity. Accord-
ing to Simon [129], such limitations are due to the concept of the bounded ration-
ality of the human mind. The satisficing strategy is a very poor decision-making
strategy, as it very frequently does not lead to expected (beneficial) results.
Another strategy followed by people when taking decisions, which is avoided
by using cognitive decision systems, is the elimination by aspect strategy [139],
which consists in determining a series of criteria and then eliminating those deci-
sions that do not meet subsequent criteria. In this strategy, an option once rejected
is no longer taken into account in subsequent considerations, even if some aspects
of it are very appealing. Due to the above defects of the human decision-making
process, cognitive UBDSS systems make use of only those human decision-
making processes that can be considered optimum solutions. These processes must
be quick in terms of the time taken to make the decision, and economical in terms
of the complexity and difficulty of cognitive operations necessary in the decision-
making process. Examples of such heuristics have been proposed in psychology,
and they comprise the following rules [39]:
•
The 'follow what is most important' rule - decisions are taken in accordance
with the rule of following certain criteria, not all of which are of the same im-
portance. The essence of this method is selecting the most important criterion,
and then comparing individual options in pairs, every time rejecting the one
whose value is lower or unknown in the light of the selected recommendation.
•
The 'follow what has worked recently' rule - decisions are taken based on the
criterion that was tried and worked in the most recent attempt of the same kind.
The presented decision-making criteria do not always allow the best decision to
be taken, and what is more often entail quite a high risk and uncertainty. This is
why the concept increasingly frequently considered to be the completely correct
description of the decision making process is the one which takes into account the
