RESEARCH PROCESS ORIENTED INTELLIGENT DECISION
SUPPORT SYSTEM FRAMEWORK
Yanjun Ji and Juanqiong Gou
School of Economic and Management, Beijing Jiaotong University, Easten Jiaoda Road, Beijing, China
Keywords: Intelligent decision support systems, Hierarchy framework, Research-oriented process.
Abstract: According to the overall development of intelligent decision support system (IDSS), we sum up the features
of some existing basic framework and lacks in dealing with problems. Based on the basic framework and
hierarchical thinking of intelligent decision support system, we make the research progress-oriented
hierarchical framework of intelligent decision support system and briefly describe the function of each
module. Finally, starting from the acquisition of knowledge system in hierarchical framework, we analyse
the knowledge’s collection and processing in intelligent decision support system, explicit of tacit knowledge
and module-chain solutions of and process knowledge.
1 INTRODUCTION
Intelligent Decision Support System (IDSS) is one
kind of Decision Support System (DSS) developed
by American scholar Bonczek and others in the 80s.
Its core idea is to combine artificial intelligence with
other relevant scientific findings to make an
intelligent decision support system, which can deal
with not only quantitative issues but also qualitative
issues.
IDSS is the combination of artificial intelligence
(AI) and decision support systems (DSS), it is
actually composite system, which integrates decision
support system (DSS) and expert system (ES). ES
simulates the ability of experts to solve problems
and to support the policy-makers to make their
decisions. DSS, which is built on the integration of
data processing and model-driven, focused on
quantitative analysis during the 1970’. Because of
having no human intelligence, lacking of knowledge
and expert support, the result of processing
uncertainty and unstructured problems was
unsatisfactory. And knowledge-based expert system
was smart enough to focus on qualitative analysis.
IDSS, which integrates ES on the basis of DSS, was
developed since then. DSS’s main parts include the
database system, model base system, human-
computer interaction system components, etc. ES is
made of knowledge, its management system and the
inference machine. The traditional way to combine
these two was to separate ES in different parts of
DSS, or the whole ES to act as a separate component
inside DSS.
IDSS, which takes advantage of both ES and
DSS, fully achieve the combination of qualitative
and quantitative. Its problem-solving ability has
greatly improved.
After years of in-depth research, intelligent
decision support system theory has been infiltrated
into the structure, problem solving and other aspects,
which has significant impact on strategy methods
and process. IDSS research has developed from
decision-making components to the components of
the comprehensive integration, from quantitative-
relied models to knowledge-based intelligent
decision-making methods, to make intelligent
decision support system theory and method more
mature. Although the theory of intelligent decision
support approach is becoming more and more
mature, but there is not too much real success,
because there is great relation between intelligence
reality and system knowledge domination, system
application framework and integration methods and
so on.
534
Ji Y. and Gou J..
RESEARCH PROCESS ORIENTED INTELLIGENT DECISION SUPPORT SYSTEM FRAMEWORK.
DOI: 10.5220/0003587105340538
In Proceedings of the 13th International Conference on Enterprise Information Systems (SSE-2011), pages 534-538
ISBN: 978-989-8425-53-9
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
2 THE HIERARCHICAL
STRUCTURE FRAMEWORK
OF INTELLIGENT DECISION
SUPPORT SYSTEM
2.1 The Hierarchical Structure
Framework of Intelligent Decision
Support System
2.1.1 The Integrated Architecture based on
Components
Since R.H. Sprgue brought up Two Libraries
(database and model base) theory of Intelligent
decision support system's basic architecture in 1980,
there has emerged the framework based X library
and the problem processing mode, and later it
developed into IDSS consisting of the problem-
solving ,human-computer interaction System, model
base systems, database systems and knowledge base
system, and knowledge systems, known as the
expert system (ES), consists of three parts including
the knowledge base, knowledge base management
system, inference engine. Its basic structure is shown
in Figure 1.
Figure 1: The three libraries basic framework of Intelligent
Decision Support.
When IDSS deal with a particular problem, it
will combines with the corresponding data into its
database, model base and knowledge base, or add a
related field database in its three-library system.
Based on this, there emerges five libraries system
structure which also includes methods and text
database. But with the increase of the number of
libraries in IDSS, the structure becomes complicated
as the function is enhanced, which make it difficult
to achieve.
2.1.2 The Structure based on the
Problem-Processing Mode
The DSS model structure based on issues-dealing
Bonczek proposed consists of knowledge system,
language system and the problem processing system.
Decision problems in which language subsystem
provides representation of decision problem and
interpretation of the decision-making structures;
knowledge subsystem is responsible for issues
related to decision-making; problem processing
subsystem gets decision problem from the language
subsystem, and through relevant knowledge
operation of knowledge subsystem, results in
rational decision-making results. The structure uses
a unified structure of knowledge- processing, which
is good to the expansion of system functions, and
can be considered as an evolutionary structure.
However, the model follows the idea of expert
system in problem solving, and can not fully reflect
the role of decision-makers in model structure and
model selection. As the knowledge required for
decision-making process are very complex, which
include factual knowledge, reasoning knowledge
and expertise knowledge in decision-making areas.
Yet such a system as a decision support system will
make the system too large and difficult to achieve,
for knowledge of this structure lack of
indiscrimination in knowledge concept and specific
issues in specific areas.
Therefore, oriented to the application for
different objects, the three library systems should be
optimized reasonably to enhance its coordination
and consistency, and it also need to be redefined the
framework in the problem solving process.
2.2 The Introduction of Hierarchical
Idea
The significant progress of General system theory
provides a new approach to intelligent decision
support system. The basic idea is: any complex
system can be viewed as a multi-model system. The
general system model commonly used has three
categories: input / output system model, target
acquisition systems model, hierarchical system
model. Therefore, the general system model can be
used as an important class of model problem in
description implementation. The introduction of
intelligent decision support method can effectively
achieve the control of the evaluation process. Based
on the classic three-library system of intelligent
decision support system, we optimize IDSS
framework referred to hierarchy, and the whole
system can be divided into the application layer,
RESEARCH PROCESS ORIENTED INTELLIGENT DECISION SUPPORT SYSTEM FRAMEWORK
535
business logic layer and data layer system, shown as
Figure 2.
Interface
Knowledge
Engineering
Process
Control
Computing
Technology
Manageme
nt
Concluding
Query
Process
Decision
Support
System
Administrati
on
KnowledgeManagementSystem
KB MB DB
Application
Layer
Business
LogicLayer
DataLayer
Figure 2: IDSS system architecture based on the Hierarchy
model.
After using Hierarchy model, the IDSS top-down
framework is divided into three layers: application
layer, business logic layer and data layer.
2.2.1 Application Layer
The application layer mainly deals with all analysis
communication between business logic layer and
users, including 4 sub-modules: interactive interface
is responsible for passing user requests and results
display, etc., it is an important channel to for users to
interact with the system. Query processor interprets
the user's natural language into executable
statements for machine searching. Decision-making
is the presenting interface for knowledge the system.
2.2.2 Business Logic Layer
Business logic layer handles all activities related to
the evaluation process, which includes five modules,
namely, knowledge engineering module, process
control modules, processing operations module,
technology management module and the conclusions
generated module. And its overall responsibility is
knowledge focus processing, process control and
results analysis.
2.2.3 Data Layer
Data layer is mainly used for knowledge, models,
data extraction, and storage and call management.
3 THE KNOWLEDGE-BASED
PROCESS IN SYSTEM WITH
HIERARCHICAL MODEL
Knowledge Systems is the core components of
intelligent decision support system. For quite some
time, the decision support system is affected by the
structure of multi-database system, it also separate
data, model, method and knowledge to build
database and management. According to different
decision tools applied, there appears several decision
support system, such as model-relied system, data-
relied system and communication-relied system and
so on. Actually, during the decision-making process,
no matter what the decision-maker gets, from fact,
rules, method to the inferential process. All those
things are essential knowledge for making a decision.
To manage that knowledge together, it will make a
concise system structure, and also a unified dealing
method, which then become a main stream of IDSS
knowledge research.
3.1 Classification of Knowledge
Knowledge involves in IDSS mainly have three
kinds: descriptive knowledge, procedural knowledge
and reasoning knowledge. Descriptive knowledge
defines various environment-related status
information for the specific decision-making,
including past, present, future and assumed data and
information. Procedural knowledge defines process,
steps and strategies in the problem solving, such as
business-running rules, action programs. Reasoning
knowledge defines how to get valid conclusions in
various possible situations. The three kinds of
knowledge are corresponding to data, models,
indicators, methods in traditional multi-database
system, and the specified relation is shown in
figure3 below:
Figure 3: Classification of Knowledge in IDSS.
We can see there are not great differences
between descriptive knowledge & reasoning
knowledge and traditional multi-database. It needs
just slight modifications for the relational database
Reasoning
Knowledge
Descriptive
Knowledge
Indicator Models
Meth
ods
Proce
ss
Rules
Data&
Information
Procedural
knowledge
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
536
in knowledge-based system. But for procedural
knowledge representation, such as the evaluation
process, it is difficult to achieve by traditional
methods, so it needs reasonable analysis and design.
3.2 Knowledge Engineering
Knowledge engineering design is based on expert
system, which is responsible for collection,
development of knowledge, access to external
information, process it by knowledge rules, store
information the form of database records to support
calling, comparing and choosing in the decision-
making process. In addition, an important function
of knowledge engineering is to interact with
knowledge workers, access to professional and
authoritative knowledge as the first reference of the
system operation control. The process is as shown in
Figure 4:
Figure 4: Schematic diagram of the knowledge
engineering process.
Ordinary users or knowledge workers and
external information are the three main sources for
knowledge engineering sources to obtain
information. Information got through these three
sources will experience knowledge extraction,
transformation and loading operations, access to
knowledge and analysis platform for analysis and
processing, filter out redundant information, and
store in knowledge base according to certain rules to
for the call.
In addition to knowledge library, the model
library can provide information about the limitations
and operating parameters on computing model. The
database can access a lot of knowledge through data
mining, data integration and other technology.
Conclusion is another important function of
knowledge engineering. It is responsible for
assessing outputs from different types of analysis
(including providing specific decision services based
on the track and analyse of user
behaviour) .Especially for the causal task; it will
screen out false or inconsistent results based on a
standard, and generates simple and profound
explanation of the model environment the user can
understand. In the process, it will use the generation
model and knowledge base, model base interaction,
deductive knowledge, analogical reasoning and
other techniques.
3.3 Dominance of Process and Other
Hidden Information
The knowledge management system is mainly
responsible for dominance of knowledge. It will
make the empirical and rational knowledge got from
a variety of information sources be structured and
formal processed, then dump them respectively, in a
particular machine readable format to knowledge
base, model base and database. Knowledge
management systems manage many knowledge
objects including data; text flow, validating model,
metadata, process, processing algorithms and the
corresponding operating software.to complete the
unified function.
Process is an important component of tacit
knowledge. A complete process mainly includes
model, index data, evaluation criteria and
operational control these four types of information.
Each model calls indicators and data in turn in
accordance with control information for processing
operations. Results operations carried out will be
determined by the evaluation criteria until the
satisfactory results are got. According to the results
of different judgments, the model will be called by
multiple times, and the model organization called in
accordance with the form of a chain is known as the
model chain. The model chain is shown as below.
Figure 5: Model chain of the process.
A model chain reflects a process, in control of all
kinds of information in the model chain running:
such as the evaluation value index which is stored in
the database as a record, then can save knowledge of
the process.
Process control controls the running process
according to a variety pre-qualification before the
model chain run , and provides the operation
interaction (including the analysis and tracking of
Indicators
•Data
•Rules
Operation contr
ol
Model1
Indicator
•Data
•Rules
Operation contr
ol
Model2
•Ect.
Ect.
Indicator
•Data
•Rules
Operation contr
ol
Modeln
RESEARCH PROCESS ORIENTED INTELLIGENT DECISION SUPPORT SYSTEM FRAMEWORK
537
user preference), that is, in the form of Process
control platform to provide the tips and guidance of
parameters ,input and output, models, knowledge in
the process of decision-making process during user
Interaction.
4 CONCLUSIONS
Intelligent decision support system is a complex
system. Using logical framework can make the
system structure more clear and the integration of
the various parts easier. Good system structure also
laid the foundation for future system
implementation. Knowledge is the core of intelligent
system. How to externalize this knowledge, which
decides the system’s degree of intelligence e.g.
process, experience, becomes increasingly
important. Based on these, the paper provides a
reference for future researchers. Next research goal
is the integration of the various components within
the hierarchy framework.
This paper was supported by “the Fundamental Research
Funds for Central Universities (2009BAG12A10-2)”
REFERENCES
Shan-Lin Yang, 2004. Machine Learning and Intelligent
Decision Support System. Beijing: Science Press.
Wang Lin, Guo’an Gao, 2000. The structure of object-
oriented model systems in IDSS. Computer &
Information Technology.
Yong Liu, 2005. Research on the Collaboration of Object-
Oriented Model & Knowledge Base. North-western
Polytechnic University’s journal.
Zhejiang Geng translated, 2009. System Analysis and
Design in a changing world. China Machine Press.
Guozhou Yang, 2005. R&D of GIS-based intelligent
decision support systems research for channel
Management. Chongqing University’s journal.
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
538