HUMAN-CENTERED META-SYNTHETIC ENGINEERING FOR
KNOWLEDGE CREATIVE SYSTEM
Cui Xia, Dai Ruwei, Li Yaodong and Zhao Mingchang
Lab of Complex Systems and Intelligence, Institute of Automation Chinese Academy of Sciences
No.95, ZhongguancunDonglu, Beijing, China
Keywords: KDD, knowledge creative, information intelligent system, human-centred, multi-agents.
Abstract: Meta-synthetic Engineering and Cyberspace for workshop of Meta-synthetic Engineering (CWME) is the
methodology for Open Complex Giant System (OCGS), proposed by distinguished scientist Dr H Tsien.
CWME synthesizes intelligence from qualitative hypothesis to quantitative studies in terms of human-
centred and human-computer cooperated manners, on a meta-level, for dealing with complexities of
Knowledge Creative System (KCS). In this paper, a new architecture for constructing knowledge creative
systems is proposed that follow on the theory of OCGS and human-centred meta-synthetic engineering,
which synthesizes data mining and multi-agent technology, as well as domain experts and users, computers
and network, relevant social components, and so on. From a broad perspective, the KCS is OCGS and
presents features such as knowledge conductive and human-centred. The system design and implementation
of KCS involves organizational factors and interaction of humans- humans, humans-computers, and
computers - computers. As a consequence, the collective intelligence emerges from the interaction network
of components in a KCS. In addition, some algorithms and tools are developed to analyze the link structure
of a KCS to distil the emergent collective wisdom on some topics.
1 INTRODUCTION
An enormous proliferation of databases in almost
every area of human endeavor has created a great
demand for new, powerful tools for turning data into
useful, task-oriented knowledge. In efforts to satisfy
this need, researchers have been exploring ideas and
methods developed in machine learning, pattern
recognition, statistical data analysis, data
visualization, neural nets, etc. These efforts have led
to the emergence of a new research area, frequently
called data mining and knowledge discovery
(Smith1998) (Ryszard 1997).
With IT and WWW development, Data mining
main topic is broader, including Association
analysis, Classification, Clustering and outlier
analysis, Sequential and spatial patterns, and time-
series analysis, Text and Web mining, data
visualization and visual data mining. For the
ultimate goal of data mining is prediction, data
mining is faced with the following Challenging
Issues:
Identifying data source for desired knowledge
(knowledge or auxiliary meta data) concerning
mining purpose.
z Data collection methods (in Web, wireless,
txt) concerning different types of data from
different environment
z Usefulness and certainty of mining results
concerning Support and confidence
z Interactive mining with different data
granularities, e.g., generalized association
rules
z Mining in data streaming environments
about look at data only once; the amount of
data is huge
z Interestingness of mining results concerning
about having to know the original
likelihood
z Evaluation of mining results i.e. How to
measure the advantage gained, Expression
of various kinds of mining results.
The above challenges in essence are to how to
judge the mining results according to original
problem, this is concerning with the tacit knowledge.
As a general rule of thumb, explicit knowledge
485
Xia C., Ruwei D., Yaodong L. and Mingchang Z. (2007).
HUMAN-CENTERED META-SYNTHETIC ENGINEERING FOR KNOWLEDGE CREATIVE SYSTEM.
In Proceedings of the Ninth International Conference on Enterprise Information Systems, pages 485-491
Copyright
c
 SciTePress

consists of anything that can be documented,
archived and codified, often with the help of IT,
especially KDD. While IT technology often
facilitate Knowledge Discovery, for Knowledge
Discovery, much harder to grasp is the concept of
tacit knowledge, or the know-how contained in
people's heads. Another, the purpose of KDD is
applied to economic domain, business domain or
others. This type system is full of uncertainty,
emergence and possibility, which belong to the open
complex giant system (OCGS). To process the
problems high related to OCGS, it need human’s
imaginary and innovation to discovery the function
structure of the problems in system. Machine can
compute in high efficiency, but can not innovate. So
complex problem solving-oriented KDD is a
knowledge creative process with human-centered
approaches.
The remainder of the paper is structured as
follows. Section 2 describes the human-centered
complex problem solving process directed by meta-
synthetic engineering. Architechture of KCS
oriented complex problems are discussed in Section
3. Section 4 describe the collective intelligence
emergence from the KCS, and a distill tool is
established. And conclusion is in Section 5.
2 HUMAN-CENTERED
COOPERATIVE COMPLEX
PROBLEM SOLVING
2.1 Characteristics of the Complex
Problem
As one of science and technology domains, systems
science takes systems as its study object from its
application to the basic theory research. Early in
1990, Chinese scientist H.S. Tsien and his
colleagues proposed a new discipline of science—
the study of open complex giant system (OCGS) and
its methodology, i.e. Meta-synthesis (meta-synthetic
engineering from the qualitative to the quantitative)
(Tsien1993, 2001).
Depending on the quantity and interactive
complexity of the subsystems and variety of
subsystems contained in the systems, system can be
divided into two large groups: simple systems and
giant systems. If the number of subsystems is
comparatively large (e.g. a hundred), such as a
manufacturing plant, it can be called a large system.
No matter which it is, small or large, such a simple
system can be studied, starting from the interaction
of the subsystems, then directly synthesizing the
dynamic function of the complete system. This can
be called the direct method. At most, a large
computer or a supercomputer is needed to process
such a system. If there are a large variety of
subsystems with hierarchical structure and complex
interrelations, then the aggregate is called a complex
giant system. As examples, there are the biological
system, human brain system, social system, etc.
what’s on the higher level is systems with human
beings as their main subsystems. For such, “open”
and “complex” have newer and broader
connotations. Here the openness can be summarized
as the following (1) system and its subsystems
exchange information with the outside world; (2) the
subsystems acquire knowledge by learning.
Moreover, the complexity of such systems can be
outlined as thus: (1) between the subsystems there
are many modes of communication; (2) subsystems
are of many varieties; (3) the subsystems have
different ways of expressing and acquiring
knowledge; (4) the structure of the subsystems
change with evoluti
There are typical OCGS, such as social system,
economic system, environment system, military
system, Internet, etc., their data usual are the study
object of KDD. Problems high related to OCGS are
full of possibility, uncertainty and emergency, which
cannot be tackled by traditional methodology or
simply putting individual techniques together. Meta-
synthetic engineering is proposed to tackle complex
problems fitting in the category of open complex
giant system (OCGS). It advocates to present an
insight of problem solving in system thinking, by the
synthesis of relevant knowledge, techniques and
intelligence, human and domain intelligence,
collective intelligence emergence on a meta-level in
analyzing, designing and implementing problem
solving-oriented Knowledge Creative System
(KCS).
Since 1990, basic research and application
research for meta-synthetic engineering have
achieved many fruit, such as intelligent system’s
meta-synthetic (Dai1995,2000), internet being a
typical OCGS(Cao2001,2003), collective wisdom
emerging from human-computer cooperated
system(Cui2003a,b), human-human interactive
models(Cui2004), human-centred cooperated system
based on meta-synthetic on, so the structure of the
system is in a state of flux. engineering design and
implement (Li2003,2004, Zhang2004), social
intelligence(Dai2004,Cui2005,Dai2006). This paper
is based on the above research result.
ICEIS 2007 - International Conference on Enterprise Information Systems
486

2.2 The Process of Human-centred
Cooperative Problem Solving on
Meta-synthetic Engineering
To solve the complex problems high related to
OCGS, the uncertainty must be taken into
consideration. That requires high efficiency of
computers and human insight into the problem, by
application of meta-synthetic engineering, which can
be derived from system thinking and thinking in
imagery supported by data information, domain’s
techniques and intelligence, human intelligence,
explicit and tacit knowledge. And then all kinds
certain factors and uncertain factor of the problem is
discovered step by step from the certain situation to
the uncertainty, illustrated by Figure 1.
Figure 1: The framework of solving complex problem full
of uncertainty.
In Figure 1, first step is to describe the problem
and its environment, then explore and recognize the
uncertain factors and possible factors. And then the
respective strategies can be established, including
strategy for problem without uncertainty, strategy
for problem considering possible factors, strategy for
problem concerning emergency and strategy being
avail of our future. With this framework, the
concrete steps following the human thinking are
established in Figure 2. There, the first is to describe
the complex problem by human’s idea and analysis
of problem, find the principal part and factors in
certain hypothesis. Secondly, focus and keys of
problem are found by experts with effective
interactions and organizations, and then build the
analysis structure of problem. Further, the
imagination of problem development is built to
recognize the key scene, select driven factors and
simulate the imagination of problem without
uncertainty. Then consider kinds of uncertainty
situations and generate strategies respectively by
evaluated evaluations standards.
The process is full of uncertainty, thinking
dynamics and thinking creativity. Many tasks are
involved such as problem analysis, dynamic
situation assessment, data information processing,
problem modelling and simulation, tactics
generation. All these should follow the human
dynamic thinking. This makes open interactive
environment very essential in every step. Thus it is
human centred.
Figure 2: The process solving complex problem based on
meta-synthetic engineering.
HUMAN-CENTERED META-SYNTHETIC ENGINEERING FOR KNOWLEDGE CREATIVE SYSTEM
487

3 ARCHITECTURE OF KCS
ORIENTED COMPLEX
PROBLEMS
3.1 KCS Inputs and Outputs
Data and information of the problem, knowledge in
human mind and in computer, and the natural rules
of human’s thought constitute the KCS inputs. On
the other hand, the research on KCS-problem
solving techniques based on Meta-synthetic
engineering generates KCS outputs. As illustrated in
figure3, this consists of knowledge in human and
computer after problem solving, constructs, models,
methods, instantiations, problem analysis open
environment, situation assessment open
environment, problem modelling and simulation
open environment.
The constructs are semantic elements that
conceptualize problems within a domain and their
solutions by means of interactions among experts,
embody experts’ explicit and tacit knowledge, and
require tools such as data mining, KDD, and
techniques of dialogue in deep thought to
implement. Models constitute a set of statements
that describe the relationships between constructs.
This needs methods such as multi-agent distributed
computing to be implemented. Instantiations are
final artefacts, limited in their scope and developed
on the basis of constructs, models and methods. In
KCS research, instantiations can precede the
complete definition of constructs, models and
methods, by having experts rely on their intuition
and experience.
Figure 3: KCS inputs and outputs.
3.2 The Abstract Multi-agent Models
for KCS
To implement the above human-centred IIS, agent
technology is very suitable. This section describes
the design and implementation of agent-based
human-centred KCS.
The KCS are distributed and should be
established on top of Internet/intranet network. In
this system, first, agent refers to man/expert and
computer in the network. Thus the abstract KCS
open model is designed to solving problems on the
basis of meta-synthetic engineering, as illustrated in
figure 4.
Here, humans and computers form a social unit.
In order to solve a problem, all social units
constitute a network-based intelligent information
system. There are collective wisdom emerging from
www society, information, data, knowledge
interactions and organizations among social units.
According to system components, there are three
type interactions and organizations: man-man
interaction and organization, man-computer
interaction and organization, and machine-machine
interaction and organization, which create a
knowledge creative environment.
Collective wisdom
www society
Problems
And Tasks
Knowledge Information Data
man
computer
orgnization
behavior
interactions
man
computer
orgnization
behavior
interactions
man
computer
orgnization
behavior
interactions
Figure 4: KCS multi-agents models on meta-synthesis.
The first involves the human knowledge (explicit
knowledge and tacit knowledge, particularly tacit
knowledge, i.e. creative thinking). In implementing
computing tasks, man’s thinking should be the core
of the computing tasks. Such a Knowledge Creative
system provides interactive techniques and
organizational mechanism for generating the space
of knowing what is in man’s head. The second type
ICEIS 2007 - International Conference on Enterprise Information Systems
488

concerns human-centred data supporting techniques.
In order to ensure the thinking to be feasible or
reasonable, KCS provides open and human-
computer-interacted data support techniques. The
third type involves distributed computing by
computer and the human-centred data supporting
techniques, where agent technology can play roles.
Figure5 introduces some details in KCS models.
3.3 Knowledge Creative Environments
3.3.1 Computer-computer Interaction and
Human-centred Data Support
Techniques
Human-centred KCS allow great flexibility to
evaluate and improve a program based on
experience in the field. They can testify user’s
dynamic thinking about a problem. Users can
change the parameter values, construct new models,
adjust the model’s structure, allow greater flexibility
to evaluate and improve a program. This type of data
support technology includes open interactive
modelling in system thinking, open interactive
visualization, and their open interactive
environments of problem analysis, situation analysis
& situation assessment, modelling and simulation.
In particular, to support this flexibility of human-
centred cooperative problem solving (
Wooldridge
1995, 1996), agent technology is a feasible tool to
design and implement computer-computer
interactions and open interactive human-centred data
supporting.
Figure 5: KCS MESSIA design framework.
To this end, KCS is designed by the method of
MESSIA (Li2003a,b), which combine the GAIA
(Wooldridge 1999,2000)and MESSAGE (Evans
2001), illustrated in Fig. 5. In system description
period, IIS consists of system organization chart,
system goal description, solving problem process
and system activities description. IIS system analysis
needs to define agent role model and interaction
model from the following five views: task/goal
view, interaction view, organization view, and agent
view. IIS system is further designed in terms of
hierarchical structures, which consist of hierarchical
agent model, hierarchical service model, and
hierarchical acquaintance model.
3.3.2 Man-man Interaction and
Organization
KCS provides interaction and organization
mechanisms for human-human to generate
knowledge (Robinson1998, Dick2003). Knowledge
is generated through interactions among groups. As
a result, KCS evolves over interactions.
Metal models, defensive routines and leaps of
abstraction (Argyris1978,1992) have a great effect
on knowledge creativity. According to the theory of
learning organization proposed by Senge et al
(Senge1990), the discussion and dialogue joined
with system thinking, self-reflection in and on
action, and balance between inquiry and advocacy
can overcome the obstacles in effective interaction,
emerge the experience of the member, change the
metal model on his own initiative, convert the
knowledge from the tacit to the explicit, from the
explicit to deeper explicit, from the explicit to the
tacit, from the tacit to deeper tacit(Nokita1995), and
emerge the collective wisdom from the global
organization of the group.
To have effective interactions among individuals,
it is necessary to apply some approaches such as
system thinking, self-reflection, balancing inquiry
and advocacy to implement self-reflective openness
discussion and dialogue for balancing inquiry and
advocacy in man-man interactions (Bohm1998).
KCS should study the mechanism and techniques for
the effective interactions.
For knowledge creativity, it also requires
learning organization. Learning organization is fused
into the coherent body of interactions. KCS is a
global organization.
Figure 6: Learning organization mechanism in KCS.
HUMAN-CENTERED META-SYNTHETIC ENGINEERING FOR KNOWLEDGE CREATIVE SYSTEM
489

In KCS, the problem solving process is
segmented into several sessions according to the
problem. Sessions may be parallel or sequential. In
each session, there are chairman and experts by
means of learning organization mechanism. Figure 6
illustrates learning organizations.
4 COLLECTIVE INTELLIGENCE
EMERGENCES
4.1 The Network of KCS
Although the process of interactions among
generalized experts is very difficult to predict at a
“local” level, it results in a network of KCS replete
with response or being responded.
Figure 7: Response and responded relations among
discussions.
Illustrated in figure7, the network of KCS is
established according to the inter-response
embedded in the discourse content, where the
discourse of expert one time is regarded as a node
V
with opinion attributes
(
)
σ
i
A
,
1
,,2,1,0 Ni K=
.
()
σ
0
A
denotes the quality attribute of an opinion,
()
σ
1
A
denotes the response quality of an opinion,
other attributes are concerning with the opinion
content by means of natural language processing.
Among these nodes, there are edges
E
with
attributes
2
,,1,0, NjH
j
K=
representing the response
embodied in discussion. Let
{
}
MkSS
k
,,2,1,0, K==
denote the generalized experts. S
0
denotes the
especial expert orienting on some certain topic---
which represent the authority opinions emerging
from www, the others represent human. Thus, the
attributed directed dynamic graph of KCS is built as
the following.
()()
(
)
⎭
⎬
⎫
⎩
⎨
⎧
=
=
=
=
t
NjHE
MkS
NiAV
G
j
k
i
2
1
,,1,0,,
,,1,0,
,,1,0,,
K
K
K
σ
(1)
Where,
HEAV ,,,
represent quantitatively the
evolution of the structure of KCS driven by the
interactions among S along with time t.
4.2 Analysis of the Link Structure
In the paper (Wooldridge and Jennings, 1996), there
is a structure analogy between KCS and www, and a
set of algorithmic tools are developed for
understanding and distilling the emergence of the
collective wisdom on the problem from the network
structures. The algorithm is overviewed here.
With hypothesis that opinion quality attribute
0
A
and the opinion response quality
1
A
exhibit what
could be called a mutually reinforcing relationship,
i.e. positive feedback, the algorithm is established.
This positive feedback is break by an iterative
method in following.
(
)
()
()
∑
∑
→
→
=
=
ji
ij
VV
ji
VV
ji
VAVA
VAVA
01
10
)(
(2)
The updating operations are performed for all
discourse of each one at every time, and the process
is repeated (normalizing the attribute value after
iteration). The discourses with the larger
authoritative attribute value are recorded, which are
the emergent representations of the emergent
intellective communities in the global level. For
example, in figure 8, the redder node has the more
authorities with larger authority value.
Figure 8: The distill results by analyze Analysis of the link
structure, the redder node the more authority.
5 CONCLUSION AND GREAT
APPLICATION
With the viewpoint of meta-synthetic engineering,
this paper analyzed the human-centred cooperative
problem solving process, and proposed a new
framework for understanding, analyzing, designing
and implementing multi-agent-based Knowledge
Creative System.
This KCS generates the models, methods,
instantiations, knowledge in both human and
computer, and the open interactive environments for
human thinking, analyzing, modelling and
simulating. These open interactive environments
require effective Interactions and organization for
man-man, man-computer, and computer-computer in
ICEIS 2007 - International Conference on Enterprise Information Systems
490

KCS. Further, interactions lead to the network of
KCS which the collective intelligence emerge from
in the system whole level. Moreover, an effective
distil tool is developed.
Since 1999, human centred metasynthetic
engineering of KCS is implemented and applied in
solving problem high related to economic system,
environment system and military system. Our further
work is on human-centred data and information
processing.
ACKNOWLEDGEMENTS
This work present some of results accumulated in
long-term research on Meta-synthesis of Intelligent
System. This involves large grants such as National
Natural Science Foundation (1999-2005,
No.79990580). Thanks also go to relevant
collaborative organizations such as Tsinghua
University, Shanghai Jiao Tong University, Xi’an
Jiao Tong University, Institute of Automation in
Chinese Science Academy and institute of system
science and math in Chinese Science Academy.
REFERENCES
Ryszard S. Michalski and Kenneth A. Kaufman, Data
Mining and Knowledge Discovery: A Review of Issues
and a Multistrategy Approach, 1997, Chapter 2 of
Machine Learning and Data Mining: Methods and
Applications, edited by R.S. Michalski, I. Bratko and
M. Kubat, John Wiley & Sons publishers, 1997 1-42
Mitchell, David, Data Mining Methods for Knowledge
Discovery.(Review), July, 2000, http://findarticles.com
/p/articles/ mi_hb3294/is_200007/ai_n8003182
H.S. Tsien (Qian XueSen), Yu Jingyuan, Dai Ruwei, a
new discipline of science---the study of open complex
giant system and its methodology, Chinese J. Of
System Engineering & Electronics, Vol.4, No.2, 1993,
pp.2-12
H.S. Tsien (Qian XueSen), system sciences, Shan Xi
Science and Technology Publishing House, China，
2001, 66-67,38-41，222-227.
Dai ruwei, Wang jue, Tian jie, Meta-synthetic engineering
of intelligent system (in Chinese), ZheJiang Science
and Technology Press in China, 1995.
Dai ruwei, metasynthetic engineering, Metallurgical
Industry Automation，2000，1，1-6
L B.Cao, Dai ruwei, Internet——one open complex giant
system, Chinese sciences(E) 2003, 33(4): 289-296.
L B Cao, Dai Ruwei, information system based on meta-
synthetic engineering----Internet, Journal of pattern
recognition and artificial intelligence, 2001, 14(1):1-8
Cui xia, Dai ruwei, Li yaodong, The Emergence of
Collective Wisdom in the Hall for Workshop of
Metasynthetic Engineering, Journal of System
Simulation, 2003(a),14(1):146-153
Cui xia, Modeling the Hall for Workshop of Metasynthetic
Engineering Based on WWW, Ph.D Dissertation,
Institute of Automation, CAS, Beijing, China, 2003(b)
Cui xia, Intelligence Emergence by means of Human-
Computer Cooperation, Dissertation for Post Doctor,
Institute of Automation, CAS, Beijing, China, 2005
Cui Xia, Li Yaodong, Dai Ruwei, A Effective Dialogue
Model of the Interactions among Experts Based on
Learning Organization in HWME, Journal of
Management Sciences in China 2004, 7(2):80-87
Li yaodong, Study on the Design and Implementation of
Hall for Workshop of Metasynthetic Engineering, PhD
Dissertation, Institute of Automation, CAS, Beijing,
China, 2003
Li yaodong, Cui xia, Dai ruwei. The Framework, Design
& Implementation of Hall for Workshop of Meta-
Synthetic Engineering, Complex systems and
complexity science, 2004, 1(1): 27-32.
Zhang jiacai, Design and Implementation of Hall for
Workshop of Metasynthetic Engineering, PhD
Dissertation, Institute of Automation, CAS, Beijing,
China , 2004
Dai Ru wei, Man-Computer Cooperative Intelligent
Science and Intelligent Technology, 2004, 6(5): 24-28.
Dai ruwei, Social intelligence [M], Shanhai,Jiao Tong
University Press, 2006.
Wooldridge Michael, Jennings Nicholas R.
Formalizing
the cooperative problem solving process. In: Readings
in Agents, 1996, 430 – 440
Wooldridge Michael, Intelligent agents, Multi-agent
Systems: A Modern Approach to Distributed Artificial
Intelligence,1995
Li Y, Cui X, Dai R. MESSIA: Combining MESSAGE with
GAIA. The 4th International Conference on Control
and Automation, 2003(a).
Li Y, Cui X, Dai R. Agent-Orientated Analysis and Design
for a New Type of Group Decision Support System.
The 7th World Multiconference on Systemics,
Cybernetics and Informatics, 2003(b)
Argyris, C., “On Organizational Learning”, 1992.
Argyris, C, Schön, D.A., “Organizational learning: a
theory of action perspective” Addison Wesley, 1978.
Peter M. Senge, The Fifth Discipline: The Art & Practice
of The Learning Organization (New York: Currency
Doubleday, 1990.
Nonaka, I., Takeuchi, H., “The knowledge-Creating
Company”, Oxford University Press, New York,
1995.pp.viii-x, 62-72.
David Bohm, “On Creativity" Published by Routledge,
London 1998
HUMAN-CENTERED META-SYNTHETIC ENGINEERING FOR KNOWLEDGE CREATIVE SYSTEM
491