KNOWLEDGE SYNTHESIS

KNOWLEDGE SYNTHESIS
J. F. Gu
Academy of Mathematics and Systems Science, Chinese Academy of Sciences
jfgu@amss.ac.cn
The most common knowledge can be acknowledged by most of people and used in the most of the cases and times. But sometime because of the ways of getting the data, information and knowledge by using different sources or inferring the knowledge by using different mechanisms in the most case people will use different knowledge to express their own thoughts at first, only through discussing and even debating people may use synthesis to reach consensus, it means that some knowledge only may be acknowledged through the synthesis of thoughts, Then during the implementation people may put the knowledge into practice also have to use the synthesis of actions. Here we wish introduce two methodologies: Meta-synthesis system approach and Wuli-Shili-Renli system approach. The first one will be useful for synthesis of thoughts, and the second one will be useful for synthesis of actions. In this paper we also will introduce expert mining to help in digging deep thought from individual expert or group of experts. Finally some case studies in the fields of economic, social and human body systems will be described.
1. Introduction
Nonaka had proposed the concept of knowledge synthesis and recently he also mentioned the synthesis of thought and synthesis of action (Nonaka et al.2008).Because of the ways of getting the data, information and knowledge by using different sources or inferring the knowledge by using different mechanisms in the most case people will use different knowledge to express their own thoughts at first, only through discussing and even debating people may use synthesis to reach consensus, it means that some knowledge only may be acknowledged through the synthesis of thoughts. Then during the implementation people may put the knowledge into practice, they have to use the synthesis of actions because of the understanding differently and benefit delegating from different authorities. Here we wish introduce two kinds of system methodologies: Meta-synthesis system approach by Qian et al (Qian et al, 1990; Gu et al, 2007a) and Wuli-Shili-Renli system approach by Gu and Zhu (Gu and Tang, 2006). The first one will be useful for synthesis of thoughts, and the second one will be useful for synthesis of actions. Certainly there exist other system approaches, such as Shinayakana system approach by Sawaragi et al and i-system by Nakamori may help us to deal with the knowledge synthesis and action synthesis, but here we will not introduce them because of limited space (Nakamori and Wierzbicki, 2005). We also will introduce expert mining to help to dig deep thought from individual expert or group of experts during synthesizing thoughts and actions. Finally some case studies in the fields of economic, social and human body systems will be described.


2. Meta-synthesis system approach
Meta-synthesis System Approach (MSA) is a Chinese system approach for solving problems related to the open, complex giant systems. MSA stands for combining the data, information, knowledge, model, expert experience and wisdom.
Data- Information-Knowledge-Model-Experience-Wisdom (DIKMEW)
MSA helps people to utilize, discover and create knowledge combining with the computer (Qian et al, 1990).
There are three kinds of meta-synthesis (see Fig. 1):
Qualitative meta-synthesis
Qualitative and quantitative meta-synthesis
From qualitative to quantitative meta-synthesis


Figure 1 Contents of meta-synthesis by Tang

2.1. Flowchart of Meta-synthesis approach (Gu et al, 2007)
The flowchart of MSA approach we developed during running NSFC major project roughly as follows
Meeting I-Analysis- Meeting II（in short M-A-M'）
or
Synchronous-Asynchronous-Synchronous
Here the first meeting( Meeting I) mainly for synthesizing the ideas and scenarios existed in experts, during the analysis stage usually we will make deep investigation and construct a series of models, the second meeting (Meeting II) we will invite not only experts, but the managers and leaders who may make the decision.
2.2. Meta-synthesis and knowledge science (Gu, 2001a)
Knowledge science helps the conversion from tacit knowledge to explicit knowledge; acquisition of collective knowledge; promotion of creating new knowledge. Meta-synthesis approach is useful for synthesizing the individual knowledge and group knowledge.

3. Wuli-Shili-Renli (WSR) system approach (Gu and Tang, 2006)
For dealing with the complex system problems the three aspects of Wuli, Shili and Renli should be considered. Here Wuli means knowing and studying the objective existence, Shili means understanding and modeling the mechanisms in the universe, Renli means compromising and coordinating the human relations. From different three li’s we will use different sciences:
from Wuli- Natural science, Technical science:
from Shili-Management science, Systems science:
from Renli- Social science and Humanity science.
3.1. Running Renli
From Renli side we will take following five aspects into consideration:
Logical thinking-Emotion-Benefit-Morality-Power
For five aspects we should use different methods and thinking:
Convince people by reasoning (Hard system thinking);
Touch people by emotion (Soft system thinking);
Convince people by benefit (Soft system thinking- Game theory);
Educate people by moral (Education);
Force people by power (Critical system thinking-Coercive, law, prison, war).
3.2. WSR working principles
We often use following working principles:
Synthesis, integration, operational, iterative.
We stand for
Combination between qualitative and quantitative methods;
Combination of Human and Computer;
Interpersonal Coordination (Knowing Coordination, Relation Coordination, Emotion Coordination, Benefit Coordination);
Coordination between Ren and Wu, Shi and Wu, Shi and Ren
3.3. Working process for WSR approach
Generally we will follow listed below stages in the working process for WSR, but not necessary strictly:

Understanding desires;
Formulating objectives;
Investigation and analysis;
Constructing policies;
Selecting alternatives;
Implementing recommendation;
Coordinating relations.

3.4. WSR for synthesis of actions
Since for solving complex problems WSR approach considers the Wuli aspect-keeping the law of nature, respect the basic law of natural science; Shili aspect- keeping the law of efficiency, respect the basic law of management science and systems engineering; Renli aspect- keeping the law of effectiveness, respect the science of humanity and society. Considering all these three aspects helps people in synthesizing actions.

4. Expert mining (Gu, 2001b; Gu, 2004a; Gu 2004b; Gu, 2006)
Expert mining is a new emergent theory and technique, which is different with data mining, text mining, web mining and expert system. It is applied for collecting the ideas, experiences, knowledge and wisdom from experts. The source for mining mainly comes from live experts. Expert mining wishes to mine the experience and knowledge existed already from one hand, and the practical experience and wisdom by experts during solving and implementing real problems from the other hand.
4.1. From data mining to expert mining
There are four lines lead to expert mining:
(1)From data mining to expert mining:
(2)From web mining to expertise oriented search:
(3)Ontology-based approach to expert mining
(4)From synthesizing expert’s opinion to expert mining
4.2. Basic concepts for expert mining
Here by the expert we may mean four types of people:
(1) General people who mainly look as sample in a large population amounts to 103- 106;
(2) Expert who contributes deep knowledge and some insights in some fields related to our problem amounts to 102-103
(3) Master who has a deep and wide knowledge and deep insights in a wider domain amounts to 10-102:
(4) Guru who has deep knowledge, insights and guidance
Expert mining is a method and tool for collecting, storing, analyzing opinion, thoughts, experiences, knowledge and wisdom from the experts and extracting useful and innovative knowledge and idea by using IT technology, computing technology, human-computer interaction and group discussion.
4.3. Meta-synthesis of Opinions (Gu, 2006)
1) Meta-synthesis of Opinions by text
Simple survey (narrative); Meta-analysis; Qualitative Meta-synthesis
2) Meta-synthesis of Opinions by meeting
Types of meeting; Ba, Facilitation, Mediation; DMTMC-system;
3) Meta-synthesis of Opinions by interview deeply
Psychology mining
5. Meta-synthesis knowledge systems (Gu, 2008)
Here we may list three examples for Meta-synthesis knowledge system (Gu, 2008):
(1)Economical System (Data +Information +Model+ Knowledge+ Experience)
(2)Social System (Data + Public Opinion + Mechanism (Social Physics) + Model + Experience + psychology)
(3)Human body System (Text + Mining + Experience+ Theory)
6. Case studies 1: Economical system
6.1. Forecasting GDP growth rate in China (JAIST case, 2003) (Gu and Tang, 2003)
6.2. Forecasting GDP growth rate in China under the impact of SARS (IIASA case, 2004) (Gu et al, 2007a)
We run these two tests in order to collect data, information and synthesize the opinions from experts with different knowledge background based on the careful calculation derived from models integration
7. Case studies 2: Social harmony system
7.1. MBA Test-discussion about some social harmony problems (MBA class, Graduate University, CAS, 2006.6.27-7.13) (Gu et al., 2007b)
In this test we invited around 40 master students to discuss 6 topics related to the social harmony problems. Before discussion we required all students to prepare their own opinions with their collected information in advance.
We divided all students into 6 groups. In each group we assigned one or two facilitators who prepare the necessary questions, knowledge and useful tool and methods for discussing deeply
Here we take topic II Housing problem as example, where we applied the PathMaker, GAE and complex network to help students to analyze and synthesize the expert opinions during the process of discussion (see Figures 2-5 ).
7.2. Taxi driver behaviors (CASCISNASS case, 2007) (Liu et al., 2007)
Because of growth of oil price taxi drivers organized strikes in some cities often, so we had used multi-agent simulation to analyze the driver behaviors under the possible situations in different oil prices, while local government takes different policies. The simulation might give some useful forecasting and conclusions for decision makers. In this case studies we also took the psychological factor into consideration. Some real
social investigation on drivers was made in providing basic data, rule and knowledge in simulation.


Fig. 2. Discussion records on the computer screen


Fig. 3. Cause-effect charts for analyzing the house price


Fig. 4. GAE－Interface for recording the discussion


Fig. 5. Housing problem –by network analysis


Fig. 6. ADVISE system structure

7.3. ADVISE system (US-DHS) (SANDIA, 2005)
This system had been designed for finding the terrorist by using the different source of data, text and web. The mainly tool for analysis is to construct the semantic network and knowledge management. It is interesting here the experts are the terrorists who had expressed, communicated each with other explicitly or tacitly (mainly). Through the network analysis analysts-experts wish to find terrorists-experts’ thoughts, connections, organization and activities. This is a large information and knowledge management system (see Figure 6). Although it had been stopped by US congress because of privacy problem, but its academic idea and design was much more advanced. This is a good example as the experts delegating the government mining the thoughts and links behind of the experts delegating terrorists.
8. Case studies 3: Traditional Chinese Medicine knowledge system
From 2006 we had participated into a large project under Ten-five year state plan supported to S&T supported by Ministry of Science and Technology, then continued to join similar project under Eleven-Five year state plan supported to S&T. The purpose of this large project was designed for collecting and maintaining the idea, experiences, knowledge and wisdom from 100 elder and famous masters in TCM. Here the experts are very famous and elder TCM doctors. We wish mine each of them the main ideas and experiences and wish also to find their collective experiences and thoughts.

8.1. TCM knowledge master mining (Tang et al. 2007a; Tang et al. 2007b)
There are some issues on TCM knowledge delivery.
1)TCM master knowledge has explicit and tacit types, we should transfer the tacit to explicit; 2) Different masters knowledge come from different ways, how can we synthesize them; 3) We wish use the information technology to support these processes. For the first task we use the SECI model proposed by Nonaka, for the second and third tasks we use the TCM Master Miner (see Figures 7 and 8).

Fig. 7. Knowledge conversion and synthesis by Tang X.J, Zhang W, Wang Z





















Fig. 8. An example from 8 master knowledge by Tang X.J, Zhang W, Wang Z

8.2. Traditional Chinese Medicine knowledge system (XYH-ISS case) (Song and Gu, 2009a, 2009b; Gu et al., 2007c)　
We use a lot of miming technologies, such as data mining, text mining and expert mining to collect the idea and thoughts in the whole process for TCM diagnosing and design a framework for synthesizing the existed, live knowledge（see Figure 9）.
　
Fig. 9．Meta-Synthesis Intelligence Mining Framework
9. Case studies 4: Social technical system
In 1995 there was a project related to preparing the diagram of standard system for the commercial accommodations and facilities in China, sponsored by State Committee of Science & Technology of China and Ministry of Internal Trade. Originally this project was done by institute of commercial machinery of Beijing only. The project had faced some trouble between the members and the leader, the leader of this project fond that the knowledge of the old members of project was not enough, he assumed that they have to add the knowledge of system engineering, so he invited us, institute of Systems Science, CAS to join this project, so from view of knowledge synthesis he was right. And from this aspect we really gave them a lot of helps in synthesizing different knowledge within different domains, different experts and users, we used the usual narrative review on many documentation, face-to-face expert investigation, brainstorming, Delphi method, cluster analysis and finally the synthesis with expert knowledge. But when we wish synthesize the actions we really met a lot of
troubles, then we used Wuli-Shili-Renli approach to solve them. We had used the CATWOE proposed by Checkland to analyze many problems related to the human relation, used some compromise between the leaders and members, members and members (Gu et al.,1997).
References
Gu J.F., Tang X.J. et al. (1997) “Wuli-Shili-Renli approach to preparing the diagram of standard system for the commercial accommodations and facilities in China”, in Wilby J. and Zhu Z. eds. The third UK-China-Japan workshop on System methodology, The University of Hull
Gu J.F. (2001a) Meta-synthesis knowledge system, Institute of Systems Science, Academy of Mathematics and Systems Sciences, Chinese Academy of Sciences, Research Report AMSS-2001-7, January 18, 2001
Gu J.F. (2001b) “On synthesizing the opinions-how can we reach a consensus”, J. Systems Engineering, Vol.16, No.5, pp340-348 (in Chinese)
Gu J.F. and Tang X.J. (2003) “A test on Meta-Synthesis system approach to forecasting the GDP growth rate in China”, in the Proceedings of 47th Annual Conference of the International Society for the Systems Sciences (Wiley, J. & Allen, J. K. eds.), R093.
Gu J.F. (2004a) “How to synthesize experts opinions-building consensus from different perspectives”, Proceedings, the fifth International Symposium on Knowledge and Systems sciences, JAIST,
291-295
Gu J.F. (2004b) “Meta-synthesis System Approach and Study on Consensus”, CASDMKM, 2004, 7, 13, Beijing
Gu J.F. (2006) “Synthesis of expert’s opinions”, DUT-Workshop on Knowledge Creation, 2006-3.27-28
Gu J.F. and Tang X.J. (2006), Wuli-Shili-Renli System approach: Theory and Applications, Shanghai Science and Technology Education Publisher, 2006 (in Chinese)
Gu J.F., Wang H.C. and Tang X.J., (2007a) Meta-synthesis method system and systematology research, Science Press. (in Chinese)
Gu J.F., Liu Y.J. and Song W.Q. (2007b) “A scientific discussion test on some social harmony problems”, Proceedings of the 51st meeting of the International Society for the Systems Sciences ISSS2007, August 5-10, Tokyo, 2007-56
Gu J.F., Liu Y.J., Song W.Q. and Zhu Z.X. (2007c) “Systems Science and Traditional Chinese Medicine”, In Nakamori Y., Wang Z.T. and Gu J.F. (Ed.) Proceedings of KSS’2007: The Eighth International Symposium on Knowledge and Systems Sciences: November 5-7, 2007:43-51, JAIST Press
Gu J.F. (2008) “Meta-synthesis Knowledge System: Basics and Practice”, March 10, 2008, DUT Workshop, Dalian
Gu J.F., Song W.Q., Zhu Z.X., Gao R. and Liu Y.J. (2008) “Expert Mining and TCM Knowledge”, KSS2008 December 11-13, 2008, Guangzhou
Liu Y.J., Niu W.Y. and Gu J.F. (2007) “Exploring Computational Scheme of Complex Problem Solving Based on Meta-Synthesis Approach”, International Conference on Computational Science (ICCS 2007) pt.4; 20070527-30
Nakamori Y. and Wierzbicki A. (2005), “Systems for integrating and creating knowledge”,
in Gu J.F. and Chroust G. eds. IFSR2005, 383-385
Nonaka I., Toyama R. and Hirata T. (2008), Managing flow: A process theory of the knowledge-based firm, Palgrave Macmillan.
Qian X. S., Yu J. Y. and Dai R. W. (1990), “A new discipline of science- The study of open complex
giant system and its methodology”, Nature Magazine, Vol. 13, pp3-10, (in Chinese)
SANDIA Report (2005) Data sciences technology for homeland security information management and knowledge discovery, SANDIA Report SAND2004-6648, 2005 Jan., http://www.doe.gov/bridge
Song W.Q. and Gu J.F. (2009 a) “Application of MSIM Platform in TCM Masters' Experience Mining”, International Conference on Industrial and Information Systems, .320-323,
Song W.Q. and Gu J.F. (2009 b) ＂Application of Information Visualization Technologies in Masters’ Experience Mining，Proceedings of 20th International Conference, MCDM, Chengdu/Jiuzhaigou, China, June 21-26, 2009.，198-201
Tang X. J., Zhang N. and Wang Z. (2007a) “TCM Masters Miner for Knowledge Transfer”, in Proceedings of IEEE SMC'2007, Montreal, October 7-10, 2007, 3541-3546
Tang X. J., Zhang N. and Wang. Z. (2007b) “Exploration of TCM Masters Knowledge Mining”, in Computational Science –ICCS2007 (proceedings, Part IV, Y. Shi, et al eds.), Lecture Notes in Computer Science, Vol. 4490, Springer-Verlag, 2007, 35-42.