![](bg3.png)
problem space, identify each data item and
information component (variable) for the nodes, the
characteristics of the information components, and
the relationships among the nodes. Finally, a
conceptual database is used to store and retrieve the
data collection gathered from the experts. The
relationships among variables and rules that apply to
problem-solving activities are described as a set of
knowledge and form a conceptual knowledge base.
Also, some researchers have tried to represent
huge data sets graphically. Computer hard disk
usage is represented by development of TreeMaps,
where huge data sets are involved and compressed
(Shneiderman, 1992). Also, a punctuation graph is
used to represent a technical document to allow the
writer to detect potentially overly complex
sentences, as well as to recognize familiar patterns
(Perlman, 1983). The approaches using Bayesian
networks or statistical graphs suffer from the
limitation that exists in many business domains; the
relationships among data are very complicated and
cannot be presented with geometric structures such
as hierarchies, linear, or networks. In other words,
due to the nature of non-geometric data or non-
spatial data, there is no obvious physical model that
can be used to represent the data that humans can
understand objectively.
Furthermore, in most business and management
domains, problem-solving is overwhelming because
of the large amount of complicated data, multiple
complex relationships among data, and the
negotiability of the constraints. Thus, in such
systems including the data with complicated
structure, it is difficult to construct a decision graph.
Furthermore, general purpose representations are not
easy to apply to a specific domain due to the
complexity of data in different domains and
sophisticated underlying functionality.
Another limitation of general statistical graphs,
is that the data to be graphed must have controllable
size or dimensions. In other words, the statistical
approaches based on decision graphs cannot
represent high-dimensional, large-sized, non-
geometric-based data for decision-making support.
This research paper focuses on developing a
research strategy for building a statistical model able
to be used for non-geometric data that are massive in
both size and dimensionality to help decision makers
eventually to improve problem-solving performance
or work alone instead of a group of experts. We will
then apply the proposed statistical model to concrete
a realistic domain to verify the effectiveness of the
model. However, the proposed statistical model
itself is domain-independent. It indicates the
procedure of a decision support system as an
automatic machine translation system which first
maps receiving questions from the user-side into an
answer from the desk-side. The final output of the
proposed model is a set of most probable answers
offering the desk staff supporting the entire human
problem-solving process in a specific business
domain. By employing a nearest neighbor (NN)
classifier the best answer obtained from the
statistical model can be chosen as an exact answer
from the desk-side.
3 STATISTICAL MODELING OF
THE DECISIONS
A simple view of decision making is that it is a
problem of a choice among several alternatives. A
somewhat more sophisticated view includes the
process of constructing the alternatives, i.e. given a
problem statement, developing a list of options. A
complete picture includes a search for opportunities
of decisions, i.e. discovering that there is a decision
to be made. For instance, a manager of a company
may face a choice in which the options are clear, e.g.
the choice of a supplier from among all existing
suppliers. There are a lot of anecdotal and some
empirical evidence that structuring decision
problems and identifying creative decision
alternatives determine the ultimate quality of
decisions. Decision support systems aim mainly at
this broadest type of decision making, and in
addition to supporting choice, they aid in modeling
and analyzing systems, such as complex
organizations, identifying decision opportunities,
and structuring decision problems.
In other words a decision support system can be
simplified to a machine translation (MT) system
translating a source sentence to a target sentence.
Machine translation is a sub-field of
computational linguistics, investigating the use of
computer software to translate text or speech from
one natural language to another. Simple machine
translation methods perform simple substitution of
words in one natural language for words in another.
More complex translations may be attempted, by
using corpus techniques in order to allow better
handling of differences in linguistic typology, phrase
recognition, and translation of idioms, as well as the
isolation of anomalies. In order to improve accuracy
of the MT methods, some research groups allow for
customization by domain or profession (such as
ICEIS 2008 - International Conference on Enterprise Information Systems
526