Si

WD

a

ARRAA

KKCFDL

1

iiefi(mwsiTvcaaasitpa

n

1d

Applied Soft Computing 12 (2012) 527–535

Contents lists available at SciVerse ScienceDirect

Applied Soft Computing

j ourna l ho mepage: www.elsev ier .com/ locate /asoc

egmenting critical factors for successful knowledge managementmplementation using the fuzzy DEMATEL method

ei-Wen Wu ∗

epartment of International Trade, Ta Hwa Institute of Technology, 1.Ta Hwa Road, Chiung-Lin, Hsin-Chu 307, Taiwan

r t i c l e i n f o

rticle history:eceived 18 May 2008eceived in revised form 5 June 2011ccepted 14 August 2011vailable online 22 August 2011

a b s t r a c t

Knowledge is a key source of sustainable competitive advantage. In response to increasingly drasticand competitive environments, many organizations wish to better utilize and manage knowledge forbusiness success. For the purpose to execute formal knowledge management (KM) effectively, someworks have suggested several critical factors of KM implementations. However, in a strategic view, sucha list of critical factors must be further honed to increase practical usefulness, as not all critical factors

eywords:nowledge management (KM)ritical factorsuzzy set theoryecision Making Trial and Evaluation

necessarily share the same importance. Moreover, assessing the importance of critical factors inevitablyinvolves the vagueness of human judgment. Hence, this study presents a favorable method combiningfuzzy set theory and the Decision Making Trial and Evaluation Laboratory (DEMATEL) method to segmentthe critical factors for successful KM implementations. Also, an empirical study is presented to illustratethe proposed method and to demonstrate its usefulness.

aboratory (DEMATEL)

. Introduction

In Taiwan, many firms recognize that utilizing and manag-ng corporate knowledge provides the competitive advantage andmproved performance, and try to employ a variety of ways tonhance their rate of knowledge creation and utilization. Somerms manage knowledge with formal knowledge managementKM) initiatives and structures, while other organizations do indeed

anage knowledge informally as part of their normal activitiesithout the use of the terminology and concepts of formal KM

tructures [20]. Knowledge has the ability to utilize information andnfluence decisions, as well as the capability to act effectively [2].he power of knowledge is a very important resource for preservingaluable heritage, learning new things, solving problems, creatingore competences, and initiating new situations for both individualnd organizations [32]. Therefore, numerous firms desire to betterctivate and leverage the knowledge for achieving value creationnd business success. In order to implement the KM effectively,ome creditable works have provided several critical factors of KMmplementation [38,53], involving business needs, KM purposes,op management support, technology, communication, culture andeople, sharing knowledge, incentives, time, measurement, cost,

nd so on.

However, in a strategic view, those critical factors are all sig-ificant but not necessarily to implement at the same time. Even

∗ Tel.: +886 3 5927700x2902; fax: +886 3 5925715.E-mail address: itmike@thit.edu.tw

568-4946/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.asoc.2011.08.008

© 2011 Elsevier B.V. All rights reserved.

a same critical factor may be differently important to individualfirm with the varied priorities; due to each organization has itsown purposes, strategies, conditions of resources, and capabilitiesin KM implementation. Especially, it is hard to obviate the possi-bility of the causal relationship within those critical factors. If thekind of causal relationship can be profoundly disclosed, the criti-cal factors are able to be well prioritized and segmented into somemeaningful groups. Hence firms can properly adjust the importanceof critical factors according to the strategic needs of different KMphases. A list of critical factors is required to be further decomposedfor higher practical usefulness. To determine the importance of crit-ical factors is a qualitative decision-making problem and inevitablyinvolves the vagueness of human judgments [33].

Thus, in terms of the critical factor segment, it is better to employan effective method which can deal with the vague judgmentsof human and model the causal relationship within critical fac-tors. The fuzzy set theory is a mathematical way which can handlevagueness in decision-making [1,68]. The Decision Making Trial andEvaluation Laboratory (DEMATEL) is a potent method which helpsfor generating a structural model and visualizing the causal rela-tionship by offering a causal diagram [11–13,18]. Hence, this studyproposes a favorable method combining the fuzzy set theory andthe DEMATEL to segment the critical factors for successful KM ini-tiatives. An empirical study is presented to illustrate the proposedmethod and to demonstrate its usefulness and validity. The rest of

this paper is organized as follows. In Section 2, some of the priorliterature related to the critical factors of KM implementation isreviewed. In Section 3, the proposed method is developed. In Sec-tion 4, an empirical study is illustrated. Finally, according to the

5 Compu

fip

2

ocriit

2

cTiirfksieiaig[

kaimibirfgtii

kebecciaesmashwbaim

28 W.-W. Wu / Applied Soft

ndings of this research, concluding remarks and suggestions areresented.

. KM implementation

Reacting to an increasingly rival business environment, numer-us organizations are emphasizing the importance of KM to createompetitive advantage, and basing the KM strategy on their uniqueesources and capabilities. For implementing the KM successfully,t is a wise way to starts with a well understanding in terms of crit-cal factors of KM implementation. The concept of knowledge andhe related critical factors are discussed below.

.1. The concept of knowledge

As [26] emphasize, competitive advantage depends on how effi-ient the firm is in building, sharing and utilizing the knowledge.here are some peculiar characteristics of knowledge, such as:t is intangible and difficult to measure, is volatile, is embodiedn agents with wills, sometimes increases through use, has wideanging impacts, often has long lead times, and can be used by dif-erent processes at the same time [63]. Especially, [31] argues thatnowledge inertia may enable or inhibit one’s ability on problemolving, which is stemming from the use of routine problem solv-ng procedures, stagnant knowledge sources, and following pastxperience or knowledge; to conquer the problem of knowledgenertia, it is necessary to update and share knowledge. Addition-lly, for knowledge to make contribution, it needs to be convertednto competencies, and competence is only important as a strate-ic resource when it is relatively distinctive to its competitors25].

Concerning the distinction between data, information, andnowledge, as [50] states, if data becomes information when theydd value, then information becomes knowledge when it addsnsight, abstraction, and better understanding. In fact, data is

ainly considered as raw numbers that once processed becomesnformation and when put in specific context, this informationecomes knowledge; the knowledge as a state of mind posits that

ndividuals expand their personal knowledge through the inputseceived from their environment [2]. According to [38], in the trans-ormation process, data is organized and structured to produceeneral information, and then the information is arranged and fil-ered to produce contextual information for specific users, nextndividuals assimilate the contextual information and transformt into knowledge.

Ref. [24] raise many types of knowledge, such as: systemicnowledge, explicit knowledge, tacit knowledge, hidden knowl-dge, and relationship knowledge. Although many categories haveeen suggested, the most frequently used distinction is tacit versusxplicit knowledge [47]. Explicit knowledge is provided by theonventional classroom instruction, which bases in data and isonverted into information; by contrast, tacit knowledge basesn practice and experience, which leads to mastery provided thewareness related to the task at hand [25]. According to [40],xplicit knowledge can be expressed in words and numbers andhared in the form of data, scientific formulae, specifications, andanuals, it can therefore be readily transmitted between individu-

ls formally and systematically; whereas tacit knowledge includesubjective insights, intuitions, and hunches, is highly personal andard to formalize, as well as is difficult to communicate or shareith others. As [39] indicates, organizational knowledge is created

y a continuous dialogue between tacit and explicit knowledge,nd there are four patterns of interaction including socialization,nternalization, externalization, and combination within a “spiral”

odel.

ting 12 (2012) 527–535

2.2. Issues of knowledge management

Organizations need to discover how to motivate their peopleto share the tacit knowledge which is the most valuable formof knowledge and is recognized as a strategic asset, though thetacit knowledge is usually very subjective and resides inside one’shead so that is difficult to communicate, comprehend and quantify[15]. The explicit knowledge is easier to be digitalized and trans-ferred, so that it can be captured and shared with others by theuse of information technology [24]. Additionally, overemphasiz-ing on explicit knowledge, especially by IT investments, may leadto a situation that companies lose their valuable tacit knowledge,whereas overemphasizing tacit knowledge may lead to a resultthat tacit knowledge on its own does not enhance innovation [24].Indeed, organization’s work with KM should focus on transposingtacit knowledge into explicit knowledge and converting individ-ual knowledge into organizational knowledge [38]. Especially, itis important to make tacit knowledge explicit at the organiza-tional level through thrust and relationship building processes [24].Further, in order to achieve sustainable competitive advantage,companies need to emphasize the total knowledge base of thecompany, i.e. the explicit-and tacit knowledge, both internally andexternally [24,26].

KM is the organizational optimization of knowledge to achieveenhanced performance, increased value, competitive advantage,and return on investment, through the use of various tools, pro-cesses, methods and techniques [28]. Also, KM is a systemic wayto manage knowledge in the organizationally specified process ofacquiring, organizing and communicating knowledge, in order toenable employees to perform more effective and productive works[2]. KM and related strategy concepts are promoted as importantcomponents for organizations to survive, because KM is regardedas a prerequisite for higher productivity and flexibility in both theprivate and the public sectors [38]. There are numbers of frame-works have developed to promote the KM implementation. Mostframeworks of the KM can be classified as prescriptive, descriptive,and a combination of the two; the prescriptive frameworks directthe ways to engage in KM activities, whereas the descriptive frame-works identify significant attributes for the success of KM initiatives[48]. According to [2], those different frameworks have many simi-larities: most of the life cycles are articulated in four phases wherethe first one is a “create” phase; and the last phase concerns theability to share and use knowledge.

The issues of KM can be studied into several aspects with dif-ferent views. Some studies deal with the topics covering entire KMactivities, such as: the successful KM process requires understand-ing the operations of the four stages [8]; KM can be split into fourseparate activities, each dealing with a particular aspect [62]; amodel of knowledge creation consists of three elements, namely,the SECI process, workplace, and the knowledge assets [41]; theknowledge manipulation activities need to be properly altered anddeployed by timely knowledge valuation [17]; and the knowledgedevelopment cycle as the process of knowledge generation, knowl-edge storage, knowledge distribution and knowledge application[2].

2.3. Successful KM implementation

In the knowledge economy, a key source of sustainable com-petitive advantage and consequent profitability relies on the wayto create, share, and utilize knowledge as a strategic resource[9,22,37,51,52]. For a solid implementation of KM, organizations

need to emphasize the knowledge base on not only explicit andtacit [24], but also internal and external [26], even individualand organizational [38]. Moreover, the frameworks of KM shouldconsider purpose/objective, knowledge, technology, learning, and

Compu

pdtr(p

mib[nipKmtmitsKpg

aTbimcemireerlIme

3

tafD

3

doddbikbs

t

W.-W. Wu / Applied Soft

eople/culture of the organization, which is a holistic and people-riven approach that considers both the knowledge cycle andhe cultural environment [48]. Successful implementation of KMequires (1) aligning the contributions of key organizational actors,2) promoting the development of knowledge networks, and (3)roviding support by delivering a purposeful message [46].

For the purpose of implementing the KM successfully, there areany critical factors required to be considered. For example, it

s important to well evaluate and select a favorable KM strategy,ecause the effective management starts with a proper strategy14]. Moreover, it is not easy to success in implementing any busi-ess activity without top management support and trust relation

n an organization, no matter how the business activity is welllanned. The KM planning is only the beginning; the successfulM implementation is the real challenge. According to [45], theain obstacles to KM implementation were: lack of ownership of

he problem, lack of time, organizational structure, senior manage-ent commitment, rewards and recognition, and an emphasis on

ndividuals rather than on teamwork. As important as awareness ofhose main obstacles is, it is also important to recognize certain keyuccess factors in KM implementation. In order to improve theseM initiatives and link them to business strategy, [35] suggest arocess-oriented knowledge management approach to bridge theap between human- and technology-oriented KM.

Understanding critical success factors will provide a hugedvantage in successful KM planning and subsequent deployment.here are several critical factors of KM implementation suggestedy some scholars and experts. For example, in order to be successful

n KM activities, [53] emphasizes that firms and their manage-ents must be entrepreneurial. Moreover, [38] suggests some

ritical elements to successfully create and implement a knowl-dge management strategy, including: purposes, support from topanagement, communication, creativity, culture and people, shar-

ng knowledge, incentives, time, evaluation, and cost. Further, [3]aises a list of KM success factors, involving strong unified lead-rship, align KM with mission and business needs, cohesive andngaged team, understand current problems and issues, collabo-ation and communication, innovation, best practices and lessonsearned, understanding and appropriate use of current technology,T infrastructure, workflow and change cycles, security, establish

etrics, reliability and integrity, accessibility and portability, cost-ffective, and interoperability.

. Methodology

For building and analyzing a model involving causal rela-ionships between complex factors, the DEMATEL is a potentnd comprehensive method. In order to extend the DEMATELor decision-making in fuzzy environments, the essentials of theEMATEL and the fuzzy set theory are discussed below.

.1. The DEMATEL method

Graph theory has grown tremendously in recent years, largelyue to the usefulness of graphs as models for computation andptimization. Applying the graph theory, we can easily visuallyiscover things inside the complex problem, because the graphisplays mathematical results with visualization clearly and unam-iguously. The DEMATEL is based on digraphs, which can separate

nvolved factors into cause group and effect group. Directed graphs,nown as digraphs, are more useful than directionless graphs,

ecause digraphs can demonstrate the directed relationships ofub-systems.

The Battelle Memorial Institute conducted a DEMATEL projecthrough its Geneva Research Centre [12,13]. The original DEMATEL

ting 12 (2012) 527–535 529

was aimed at the fragmented and antagonistic phenomena of worldsocieties and searched for integrated solutions. In recent years, theDEMATEL has become very popular in Japan [18,27,66,67], becauseit is especially pragmatic to visualize the structure of complicatedcausal relationships with digraphs. The digraph portrays a basicconcept of contextual relation among the elements of the system, inwhich the numeral represents the strength of influence. A digraphmay typically represent a communication network, or some dom-ination relation between individuals. Suppose a system contains aset of elements S = {s1, s2, . . ., sn}, and particular pair-wise relationsare determined for modeling with respect to a mathematical rela-tion R. Further, to portray the relation R as a direct-relation matrixthat is indexed equally on both dimensions by elements from theset S. Then, except the case that the number is 0 appearing in thecell (i, j), if the entry is a positive integral that has the meaning of:(1) the ordered pair (si, sj) is in the relation R, and (2) there has thesort of relation regarding that the element si causes element sj.

The DEMATEL can map out complex relationships among factorsand to identify key factors [34,56–60], which is based on digraphsthat portrays a contextual relation among the elements of the sys-tem and can be converted into a visible structural model of thesystem with respect to that relation [42]. The tangible product of theDEMATEL exercise is a structural model appearing as a “causal dia-gram” which may divide sub-systems into cause group and effectgroup. In particular, DEMATEL has the ability not only to demon-strate directed relationships of sub-systems, but also to clarify thedegree of interactions between sub-systems. Thus, toward ana-lyzing a complex system, if we wish to capture the causal–effectrelationship among sub-systems, apparently the DEMATEL is help-ful. In order to apply the DEMATEL smoothly, this study refined theversion of [11]. Essential definitions are described as below.

Definition 1. The initial direct-relation matrix Z is a n × n matrixobtained by pair-wise comparisons in terms of influences anddirections between criteria, in which Zij is denoted as the degreeto which the criterion i affects the criterion j, i.e., Z = [Zij]n×n.

Definition 2. The normalized direct-relation matrix X, i.e.,X = [Xij]n×n and 0 ≤ xij ≤ 1 can be obtained through formulas (1) and(2), in which all principal diagonal elements are equal to zero.

X = s · Z (1)

s = 1

max1≤i≤n

∑nj=1zij

, i, j = 1, 2, ..., n (2)

Definition 3. The total-relation matrix T can be acquired by usingthe formula (3), in which the I is denoted as the identity matrix.

T = X(I − X)−1 (3)

Definition 4. The sum of rows and the sum of columns are sepa-rately denoted as D and R through the formulas (4)–(6).

T = tij, i, j = 1, 2, ..., n (4)

D =n∑

j=1

tij (5)

R =n∑

i=1

tij (6)

where D and R denote the sum of rows and the sum of columns,respectively.

Definition 5. A causal diagram can be acquired by mapping thedataset of (D + R, D − R), where the horizontal axis (D + R) is made byadding D to R, and the vertical axis (D − R) is made by subtractingD from R.

5 Compu

3

g[uvtsnse

bmoieglwEuaaittts[

DXbfve�

i

a

Di

�

∀

Di

Do

Dd

Dt

�

w

30 W.-W. Wu / Applied Soft

.2. Fuzzy set theory

In the real world, many decisions involve imprecision due tooals, constraints, and possible actions are not known precisely1], judgments for decision-making are often given by crisp val-es, though crisp values are an inadequate reflection of situationalagueness. To solve this kind of imprecision problem, fuzzy setheory was first introduced by [68] as a mathematical way to repre-ent and handle vagueness in uncertainty. In fuzzy set theory, eachumber between 0 and 1 indicates a partial truth, whereas crispets correspond to binary logic: 0 or 1. Hence, fuzzy set theory canxpress and handle vague or imprecise judgments mathematically.

Generally, decision-makers usually tend to give assessmentsased on their past experiences and knowledge, and also their esti-ations are often expressed in equivocal linguistic terms. Based

n the definition of fuzzy sets, the concept of linguistic variables isntroduced to represent a language typically adopted by a humanxpert. A linguistic variable is a variable whose values, namely lin-uistic values, have the form of phrases or sentences in a naturalanguage [4,19,61]. The linguistic variable is very useful in dealing

ith situations which are described in quantitative expressions.specially, linguistic variables are used as variables whose val-es are not numbers but linguistic terms [69]. The linguistic termpproach is a convenient way for decision makers to express theirssessments. In order to efficiently resolve the ambiguity arisingn incomplete information and the fuzziness in human judgments,he use of linguistic scales is necessary and important. In practice,he linguistic values can be represented by fuzzy numbers, and theriangular fuzzy number is commonly used. This study builds onome important definitions and notations of fuzzy set theory from69] and [5]. The related definitions are as follows.

efinition 6. A fuzzy set A is a subset of a universe of discourse, which is a set of ordered pairs and is characterized by a mem-ership function �A(x) representing a mapping �A : X → [0, 1]. Theunction value of �A(x) for the fuzzy set A is called the membershipalue of x in A, which represents the degree of truth that x is anlement of the fuzzy set A. It is assumed that �A : X ∈ [0, 1], where

A(x) = 1 reveals that x completely belongs to A, while �A(x) = 0ndicates that x does not belong to the fuzzy set A.

A = {(x, �A(x))}, x ∈ X, where �A(x) is the membership functionnd X = {x} represents a collection of elements x.

efinition 7. A fuzzy set A of the universe of discourse X is convexf

A(�x1 + (1 − �)x2) ≥ min(�A(x1), �A(x2)),

x ∈ [x1, x2], where � ∈ [0, 1] (7)

efinition 8. A fuzzy set A of the universe of discourse X is normalf max �A(x) = 1

efinition 9. A fuzzy number N is a fuzzy subset in the universef discourse X, which is both convex and normal.

efinition 10. The �-cut of the fuzzy set A of the universe ofiscourse X is defined as A˛ = {x ∈ X|�A(x) ≥ ˛}, where � ∈ [0,1].

efinition 11. A triangular fuzzy number N can be defined as ariplet (l, m, r), and the membership function �N(x) is defined as:

N(x) =

⎧⎪⎪⎪⎪⎨0, x < l,(x − l)(m − l)

, l ≤ x ≤ m,

⎪⎪⎪⎪⎩(r − x)(r − m)

m ≤ x ≤ r,

0, x > r,

here l, m, and r are real numbers and l ≤ m ≤ r.

ting 12 (2012) 527–535

For achieving an effective solution of problem-solving, thegroup decision-making is important to any organization, becauseit usually impacts upon those decisions that affect organizationalperformance. Specifically, the group decision-making is the processof arriving at a consensus based upon the reaction of multiple indi-viduals, and it can facilitate the exchange of ideas and informationwhereby an acceptable judgment may be obtained [6,30].

There are several useful defuzzification methods which canbe divided into two classes by considering either the vertical orthe horizontal representation of possibility distribution [44]. Inachieving a favorable solution, the group decision-making is usu-ally important to any organizations. To deal with the problems inuncertainty, an effective fuzzy aggregation method is required. Anyfuzzy aggregation method always needs to contain a defuzzificationmethod due to the results of human judgments with fuzzy linguis-tic variables are based on TFNs. The defuzzification refers to theselection of a specific crisp element based on the output fuzzy set,which convert fuzzy numbers into crisp may score. This study isapplying the converting fuzzy data into crisp scores developed by[44], the main procedure of determining the left and right scoresby fuzzy minimum and maximum, the total score is determined asa weighted average according to the membership functions.

This study here adopts the CFCS (Converting Fuzzy data intoCrisp Scores) defuzzification method for the fuzzy aggregation pro-cedure, because the CFCS method can give a better crisp value thanthe Centroid method. The CFCS method is based on the proce-dure of determining the left and right scores by fuzzy min andfuzzy max, respectively, and the total score is determined as aweighted average according to the membership functions [42]. Letzk

ij= (lk

ij, mk

ij, rk

ij) indicate the fuzzy assessments of evaluator k (k = 1,

2,. . .,p) about the degree to which the criterion i affects the crite-rion j. To aggregate the result of these fuzzy assessments, this studyuses the CFCS method which includes five-step algorithms.

Assume X to be an arbitrary convex and bounded fuzzy num-ber. The assessed values of qualitative criteria metrics for NBSC,X = (Lxij, mxij, Rxij), i = 1,2,3,4 and j = 1,2,3. . .,7 in this study. X =(Lxij, mxij, Rxij) is TFNs, and xij presents at the left, middle and rightpositions, Lxk

ij, mxk

ij, Rxk

ij, represent overall average ratings of aspect

ith, criteria jth over kth evaluators, and xpij, p = 1, 2,. . .. . .k, is fuzzy

numbers for each evaluator. The normalization of TFNs as follows:

(1) Normalization:

xlkij =(lk

ij− min lk

ij)

�maxmin

(8)

xmkij =

(mkij

− minlkij)

�maxmin

(9)

xrkij =

(rkij

− minlkij)

�maxmin

(10)

where �maxmin = maxrk

ij− minlk

ij.

(2) Compute left (ls) and right (rs) normalized value:

xlskij =

xmkij

(1 + xmkij

− xlkij)

(11)

xrskij =

xrkij

(1 + xrkij

− xmkij)

(12)

(3) Compute total normalized crisp value:

xkij =

[xlskij(1 − xlsk

ij) + xrsk

ijxrsk

ij]

[1 − xlskij

+ xrskij]

(13)

W.-W. Wu / Applied Soft Compu

Table 1The fuzzy linguistic scale.

Linguistic terms Triangular fuzzy numbers

Very high influence (VH) (0.75,1.0,1.0)High influence (H) (0.5,0.75,1.0)Low influence (L) (0.25,0.5,0.75)

(

(

3

ttim

Deoartts

lnncfmalas

mCtfdt(

brttEa“id(wmr

Very low influence (VL) (0,0.25,0.5)No influence (No) (0,0,0.25)

4) Compute crisp values:

zkij = minlkij + xk

ij�maxmin (14)

5) Integrate crisp values:

zij = 1p

(z1ij + z2

ij + · · · + zpij) (15)

.3. The proposed method

The DEMATEL method is a highly pragmatic way to form a struc-ural model of evaluation for better decision making. To furtherhe practicality of the DEMATEL method for group decision makingn a fuzzy environment, the analytical procedure of the proposed

ethod is explained as follows:Step1: identifying the decision goal and forming a committee.

ecision making is the process of defining the decision goals, gath-ring relevant information, generating the broadest possible rangef alternatives, evaluating the alternatives for advantages and dis-dvantages, selecting the optimal alternative, and monitoring theesults to ensure that the decision goals are achieved [16,43]. Thus,he first step is to identify the decision goal. Also, it is necessaryo form a committee for gathering group knowledge for problem-olving.

Step2: developing evaluation factors and designing the fuzzyinguistic scale. In this step, it is necessary to establish sets of sig-ificant factors for evaluation. However, evaluation factors have theature of causal relationships and are usually comprised of manyomplicated aspects. To gain a structural model dividing involvedactors into cause group and effect group, the DEMATEL method

ust be used here. For dealing with the ambiguities of humanssessments, the linguistic variable “influence” is used with fiveinguistic terms [29] as {Very high, High, Low, Very low, No} thatre expressed in positive triangular fuzzy numbers (lij, mij, rij) ashown in Table 1.

Step3: acquiring and aggregating the assessments of decisionakers. To measure the relationship between evaluation factors

= {Ci|i = 1, 2, ..., n}, it is usually necessary to ask a group of expertso make assessments in terms of influences and directions betweenactors. Then, using the CFCS method, those fuzzy assessments areefuzzified and aggregated as a crisp value which is the zij. Hence,he initial direct-relation matrix Z = [zij]n × n can be obtained by Eqs.7)–(15).

Step4: establishing and analyzing the structural model. On thease of the initial direct-relation matrix Z, the normalized direct-elation matrix X can be obtained through Eqs. (1) and (2). Then, theotal-relation matrix T can be acquired by using Eq. (3). Accordingo Definitions 5 and 6, the causal diagram can be acquired throughqs. (4)–(6). The causal diagram is constructed with the horizontalxis (D + R) named “Prominence” and the vertical axis (D − R) namedRelation”. The horizontal axis “Prominence” shows how muchmportance the factor has, whereas the vertical axis “Relation” mayivide factors into cause group and effect group. Generally, if the

D − R) axis is plus, the factor belongs to the cause group; other-ise, the factor belongs to the effect group if the (D − R) axis isinus. Hence, causal diagrams can visualize the complicated causal

elationships of factors into a visible structural model, providing

ting 12 (2012) 527–535 531

valuable insight for problem-solving. Further, with the help of acausal diagram, we can make better decisions by recognizing thedifference between cause and effect factors.

4. Empirical study and discussions

Being in need of enhanced competitive advantage, most orga-nizations wish to enrich and utilize knowledge effectively. Inthis section, an empirical study shows how a high-tech companyapplied the proposed method to segment a list of critical factors fora successful KM initiative.

4.1. Problem descriptions

Case Company G is a Taiwan firm with more than USD 250 mil-lion turnover and over 1250 employees worldwide. The companyis one of the world’s leading manufacturers in the Broadband Wire-less Networking business, offering various solutions and productsranging from Wireless ADSL technology, Access Points, WirelessRouters, Client Adapters, and Built-in Modules. In order to succeedin a dynamic business environment, it is now a leading com-pany strategy to apply KM to create, share, and utilize knowledgeto increase competitive advantages. Also, Company G wanted totransform and leverage their knowledge into competitive advan-tages through formal KM implementation. However, Company Gran into trouble when making KM initiatives, because any KMinitiative needs to take into account several complex factors sys-tematically, such as: purpose; the condition of resources and theircapabilities; even the preferences of a company.

Although they recognized many critical factors in successful KMimplementation, there arose the problem (since those critical fac-tors were not equally important) of how to segment them intomeaningful groups. In order to acquire sensible segments, Com-pany G therefore set up a KM development committee consistingof the General Manager and several managers representing themarketing, financial, production, human resource, and informa-tion technology departments. The following shows how CompanyG utilized the proposed fuzzy DEMATEL method to evaluate andsegment a list of critical factors for its KM initiative.

4.2. Applications of proposed method

The committee followed the proposed method with thefour-step procedure. First, they defined the decision goals for seg-menting critical factors into significant groups in order to launchthe KM initiative successfully. In step 2, the committee built andinspected a list of critical factors which was mainly based on theworks of [38] and [3]. Those factors were: top management sup-port (C1), communication (C2), culture and people (C3), sharingknowledge (C4), incentives (C5), time (C6), trust (C7), cost (C8), per-formance measurements (C9), information technology (C10), andsecurity (C11). Also, they decided to use the fuzzy linguistic scale(Table 1) for making assessments.

In step 3, once the relationships between those factors weremeasured by the committee through the use of the fuzzy linguisticscale, the data from each individual assessment could be obtained.For example, the assessment data of the General Manager areshown in Table 2. Then, using the CFCS method to aggregate theseassessment data, the initial direct-relation matrix (Table 3) wasproduced. In step 4, based on the initial direct-relation matrix, thenormalized direct-relation matrix (Table 4) was obtained by formu-las (1) and (2). Next, the total-relation matrix (Table 5) was acquired

using formula (3). Then, using formulas (4)–(6), the causal diagram(Fig. 1) could be acquired by mapping a dataset (see Table 5) of(D + R, D − R). Looking at this causal diagram, it is clear that evalu-ation factors were visually divided into the cause group including:

532 W.-W. Wu / Applied Soft Computing 12 (2012) 527–535

Table 2The assessment data of the general manager.

C1 C2 C3 C4 C5 C6 C7 CS C9 C10 C11

C1 No VH H VH H VH VH VH VH H HC2 H No L H VL L L H H L LC3 H VH No VH VH VH VH VH VH VH HC4 L L L No VL VL VL VL L L LC5 H H H H No L L L L L HC6 VL L L L VL No No VL H H HC7 VL L L L VL No No VL H H HC8 L H L L VL VL VL No H H LC9 No VL VL VL No No No No No VL VLC10 No No VL VL No No No No VL No VLC11 No VL VL VL No No No No VL VL No

Table 3The initial direct-relation matrix.

C1 C2 C3 C4 C5 C6 C7 CS C9 C10 C11

C1 0.000 0.800 0.869 0.700 0.839 0.738 0.770 0.633 0.834 0.839 0.600C2 0.802 0.000 0.673 0.738 0.600 0.641 0.705 0.705 0.667 0.705 0.500C3 0.869 0.834 0.000 0.834 0.929 0.867 0.802 0.899 0.700 0.929 0.667C4 0.641 0.533 0.500 0.000 0.400 0.467 0.467 0.359 0.567 0.545 0.673C5 0.839 0.869 0.770 0.600 0.000 0.567 0.577 0.533 0.467 0.600 0.633C6 0.600 0.567 0.567 0.467 0.359 0.000 0.131 0.367 0.633 0.600 0.567C7 0.500 0.609 0.577 0.533 0.500 0.200 0.000 0.391 0.667 0.667 0.633C8 0.567 0.633 0.533 0.500 0.359 0.467 0.433 0.000 0.567 0.733 0.533C9 0.230 0.400 0.400 0.500 0.333 0.263 0.263 0.400 0.000 0.400 0.359C10 0.367 0.333 0.533 0.391 0.333 0.367 0.263 0.467 0.400 0.000 0.433C11 0.263 0.359 0.400 0.533 0.220 0.220 0.327 0.300 0.433 0.327 0.000

Table 4The normalized direct-relation matrix.

C1 C2 C3 C4 C5 C6 C7 CS C9 C10 C11

C1 0.000 0.096 0.104 0.084 0.101 0.089 0.092 0.076 0.100 0.101 0.072C2 0.096 0.000 0.081 0.089 0.072 0.077 0.085 0.085 0.080 0.085 0.060C3 0.104 0.100 0.000 0.100 0.112 0.104 0.096 0.108 0.084 0.112 0.080C4 0.077 0.064 0.060 0.000 0.048 0.056 0.056 0.043 0.068 0.065 0.081C5 0.101 0.104 0.092 0.072 0.000 0.068 0.069 0.064 0.056 0.072 0.076C6 0.072 0.068 0.068 0.056 0.043 0.000 0.016 0.044 0.076 0.072 0.068C1 0.060 0.073 0.069 0.064 0.060 0.024 0.000 0.047 0.080 0.080 0.076C8 0.068 0.076 0.064 0.060 0.043 0.056 0.052 0.000 0.068 0.088 0.064

Cs

4

fo

TT

C9 0.028 0.048 0.048 0.060 0.040

C10 0.044 0.040 0.064 0.047 0.040

C11 0.032 0.043 0.048 0.064 0.026

1, C2, C3, C4, C5, C6 and C7 while the effect group was composed ofuch factors as C4, C9, C10 and C11.

.3. Discussions

In this empirical study, the case Company wanted to implementormal KM in a stepwise manner, and needed to segment a listf critical factors into meaningful groups for making decision in

able 5he total-relation matrix.

C1 C2 C3 C4 C5 C6 C7

C1 0.169 0.264 0.268 0.251 0.240 0.227 0.229

C2 0.239 0.157 0.230 0.236 0.199 0.201 0.207

C3 0.279 0.282 0.188 0.278 0.261 0.253 0.243

C4 0.186 0.179 0.175 0.119 0.147 0.152 0.151

C5 0.241 0.250 0.237 0.220 0.131 0.193 0.193

C6 0.176 0.177 0.176 0.166 0.138 0.095 0.110

C7 0.174 0.190 0.185 0.182 0.160 0.125 0.101

C8 0.182 0.193 0.182 0.179 0.145 0.156 0.151

C9 0.109 0.130 0.128 0.140 0.109 0.101 0.100

C10 0.131 0.131 0.151 0.137 0.117 0.120 0.107

C11 0.108 0.121 0.124 0.139 0.093 0.092 0.104

R 1.993 2.073 2.045 2.046 1.739 1.715 1.694

0.032 0.032 0.048 0.000 0.048 0.0430.044 0.032 0.056 0.048 0.000 0.0520.026 0.039 0.036 0.052 0.039 0.000

successful KM initiatives. According to the result from this pro-posed method, several implications about business managementcan be derived as follows.

It is important to distinguish whether a critical factor belongs to

the cause group factors or the effect group. The cause group impliesthe meaning of the influencing factors, whereas the effect groupdenotes the meaning of the influenced factors. If we want to reacha high level of performance in terms of the effect group factors,

CS C9 C10 C11 D (D + R) (D − R)

0.224 0.268 0.280 0.233 2.653 4.646 0.6600.215 0.232 0.247 0.205 2.368 4.441 0.2940.264 0.268 0.305 0.255 2.876 4.921 0.8310.146 0.184 0.189 0.189 1.817 3.862 −0.2290.195 0.208 0.233 0.217 2.317 4.057 0.5780.142 0.185 0.189 0.171 1.725 3.440 0.0100.152 0.197 0.205 0.187 1.857 3.551 0.1630.108 0.188 0.214 0.177 1.875 3.680 0.0700.119 0.084 0.136 0.121 1.276 3.360 −0.8090.134 0.139 0.100 0.137 1.405 3.625 −0.8160.105 0.130 0.123 0.076 1.215 3.183 −0.753

1.805 2.085 2.221 1.968

W.-W. Wu / Applied Soft Computing 12 (2012) 527–535 533

C1

C2

C3

C4

C5

C6

C7

C8

C9C10

C11

-1.00

-0.80

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

0.00 1.00 2.0 0 3.00 4.00 5.0 0 6.00

D-R

D+R

iclrCtf(ifhtbeptc(

sgrtcoiic[ip

-0.80

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

0.80

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 D-R

D+R

C11

C7

C6

C2

C3

C8

C5

C9

C1

C4

C10

Fig. 2. The causal diagram (b).

Table 7The values of (Di + Ri)

defand (Di − Ri)

def.

(Di + Ri)def

(Di − Ri)def

C1 4.366 0.479C2 4.241 0.262C3 4.551 0.552C4 3.811 −0.163C5 3.939 0.396C6 3.434 0.119C7 3.557 0.152C8 3.632 0.080C9 3.379 −0.555

TT

Fig. 1. The causal diagram (a).

t is necessary to control and pay a great deal of attention to theause group factors beforehand. From the result of segmenting theist of critical factors, it means that successful KM implementationequires a high level of focus on the cause group (C1, C2, C3, C5, C6,7 and C8) rather than the effect group (C4, C9,C10 and C11); thoughhe cause group factors are difficult to move, while the effect groupactors are easily moved [18]. Further, through this causal diagramFig. 1.) several valuable cues can be obviously obtained for mak-ng profound decisions. For example, among these eleven criticalactors, culture and people (C3) is the most important factor by theighest (D + R) priority of 4.921. Also, it is the most influencing fac-or by the highest (D − R) priority of 0.831, but it is quite difficult toe changed. As to the information technology (C10), it is the mostasily influenced and moved factor because it has the lowest (D − R)riority of minus 0.816. Moreover, we can directly look those fac-ors scattered in the causal diagram and perceive that three keyritical factors for successful KM initiative are: culture and peopleC3), top management support (C1), and incentives (C5).

With the proposed fuzzy DEMATEL method, the case Companyuccessfully segmented a list of critical factors into expressiveroups for making decision in the KM initiative. According theesults of segmentation, it was revealed that the most crucial fac-ors are culture and people, not information technology. Althoughulture and people are not easily changed, they are the core partf promoting a successful KM initiative and the root of creat-ng sustainable competitive advantage. Knowledge does not existndependent of human experience [49]. Several studies have indi-

ated that culture and people issues are the most decisive factors7,10,15,23,45,48]. Hence, if the case Company wishes to succeedn its KM initiative, it must emphasize the importance of peo-le and to nurture a favorable culture such as an innovative and

able 6he values of Di + Ri and Di − Ri .

Ri Di

C1 (0.520, 1.413, 3.897) (0.803, 1.887, 4.578)

C2 (0.544, 1.466, 3.958) (0.671, 1.683, 4.402)

C3 (0.519, 1.444, 4.036) (0.911, 2.058, 4.686)

C4 (0.515, 1.441, 4.004) (0.412, 1.279, 3.782)

C5 (0.403, 1.239, 3.557) (0.641, 1.637, 4.338)C6 (0.394, 1.196, 3.512) (0.385, 1.205, 3.610)

C7 (0.387, 1.184, 3.505) (0.449, 1.307, 3.839)

C8 (0.419, 1.265, 3.653) (0.433, 1.311, 3.817)

C9 (0.532, 1.465, 4.025) (0.190, 0.881, 3.043)

C10 (0.605, 1.579, 4.190) (0.260, 0.983, 3.176)

C11 (0.478, 1.382, 3.922) (0.160, 0.844, 2.989)

C10 3.598 −0.675C11 3.258 −0.563

entrepreneurial culture. Finally, all KM initiatives are unique so thata segmented result may not be completely suitable for other com-panies. However, the proposed fuzzy DEMATEL method is quiteuseful in segmenting several critical factors into profound groupsfor making better decisions in a fuzzy environment.

Additionally, [36] suggest a fuzzy DEMATEL solution which isbetter than other studies that aggregating all the data of the expertsright after obtaining the initial direct-relation fuzzy matrix. Hence,it is interesting to conduct comparison with the fuzzy DEMATELsolution suggested by [36]. As a result, we can obtain the causaldiagram (Fig. 2) based on Tables 6 and 7. The main dissimilarity

between Figs. 1 and 2 is that the location of C9. Although these twofuzzy DEMATEL methods produce almost similar results, this doesnot mean that the fuzzy DEMATEL developed by [36] is not useful.

Di + Ri Di − Ri

(1.322, 3.300, 8.475) (−3.094, 0.474, 4.058)(1.215, 3.149, 8.361) (−3.288, 0.216, 3.858)(1.430, 3.501, 8.722) (−3.125, 0.614, 4.167)(0.927, 2.719, 7.786) (−3.592, −0.162, 3.266)(1.044, 2.876, 7.895) (−3.145, 0.398, 3.935)(0.779, 2.401, 7.122) (−2.870, 0.009, 3.216)(0.836, 2.491, 7.345) (−3.120, 0.122, 3.453)(0.852, 2.576, 7.470) (−3.203, 0.045, 3.398)(0.722, 2.346, 7.068) (−3.592, −0.584, 2.510)(0.865, 2.561, 7.366) (−4.000, −0.596, 2.571)(0.638, 2.226, 6.911) (−3.661, −0.537, 2.511)

5 Compu

Tr

5

oaaimntcfD

tftbsarjbwcvrHse

ltDaimcdi

sasscamamrtttt

agpDsfs

[

[

[

[

[

[

[

[

[

[

[

[

[

[

34 W.-W. Wu / Applied Soft

hese two fuzzy DEMATEL methods can complement rather thaneplace one another in order to conduct informed analyses.

. Concluding remarks

Knowledge is the fundamental basis of competition, so thatrganizations must endeavor to enrich their knowledge resourcesnd need to design a knowledge strategy to enhance a sustain-ble competitive advantage. A successful KM initiative requiresdentifying of critical factors which guide the success of KM imple-

entation. However, all critical factors are significant, but do notecessarily share the same importance, even having causal rela-ionships between them. With a strategic view, such a list ofritical factors must be further honed for higher practical use-ulness. Rather than just simply ranking the critical factors, theEMATEL method provides a favorable solution.

The DEMATEL method is based on graph theory that enables uso project and solve problems visually, and it can divide multipleactors into cause group and effect group in order to better cap-ure causal relationships visibly, as well as convert the relationshipetween critical factors into an intelligible structural model of theystem. However, in many cases, the judgments of decision-makingre often given as crisp values, but crisp values are an inadequateeflection of the vagueness in the real world. The fact that humanudgment about preferences are often unclear and hard to estimatey exact numerical values has created the need for fuzzy set theoryhen handling problems characterized by vagueness and impre-

ision. A more sensible approach is to use, instead of numericalalues, linguistic assessments in which all assessments of crite-ia in the problem are evaluated by means of linguistic variables.ence, there is a need to extend the DEMATEL method with fuzzy

et theory and linguistic variables for decision-making in fuzzynvironments.

However, in order to handle this kind of fuzzy MCDM prob-em in terms of the critical factor segment, this study developedhe fuzzy DEMATEL method. This proposed method extends theEMATEL method by applying both linguistic variables and a fuzzyggregation method, so that it can effectively deal with vague andmprecise judgments in group decision-making. In particular, this

ethod can also successfully divide a set of complex factors into aause group and an effect group, as well as giving a visible causaliagram. Through the causal diagram, the complexity of a problem

s easier to capture, whereby profound decisions can be made.The DEMATEL has been successfully applied in a variety of fields

uch as: finding critical services [29], importance-performancenalysis [21], selecting management systems [53]; a value-createdystem of science park [35], choosing knowledge managementtrategies [64], corporate social responsibility programs choice andosts assessment [54], group decision-making [36], safety man-gement system [34], innovation policy portfolios [20], globalanagers’ competencies [65], the system failure mode and effects

nalysis [50], performance evaluation [65], municipal solid wasteanagement [10,15], and so on. Yet, apart from [10,15,55], it is

arely to use the DEMATEL for dealing with the issue of KM. Thus,his paper segments critical factors for successful KM implemen-ation using the fuzzy DEMATEL method, and successfully extendshe practical applications of fuzzy set theory and EDEMATEL intohe field of KM.

The proposed fuzzy DEMATEL method is comprehensive andpplicable to all organizations facing difficult problems that requireroup decision-making in the fuzzy environments to segment com-lex factors. As concerns this empirical study, the proposed fuzzy

EMATEL method worked smoothly in tackling the problem of

egmenting the critical factors into meaningful groups in order toacilitate the KM initiative. The result of this study indicates that auccessful KM initiative needs to highlight critical factors such as:

[

ting 12 (2012) 527–535

culture and people, top management support, incentives, commu-nication, and so on. Especially, the root causes is the culture andpeople that may influence other factors when implementing KMactivities. The finding not only offers a meaningful base to deepenthe understanding with regard to the KM initiative, but also pro-vides a clue to develop effective interventions to promote the KMimplementation with a stepwise manner. However, the study hassome limitations. First, the study only conducted a case study; thefinding should not be generalized to other enterprises. Second, itis believed that different enterprises may have different concernsabout criteria for KM implementation. In this sense, it is worth-while to perform more cases study in order to unearth new criteriafor use. Additionally, it calls for periodical diagnoses in order tograsp the dynamic KM activities with different interventions andpromotion strategies.

References

[1] R.E. Bellman, L.A. Zadeh, Decision-making in a fuzzy environment, ManagementScience 17 (4) (1970) 141–164.

[2] H. Benbya, G. Passiante, N.A. Belbaly, Corporate portal: a tool for knowledgemanagement synchronization, International Journal of Information Manage-ment 24 (3) (2004) 201–220.

[3] C. Bixler, Knowledge management and the learning organization converge, KMWorld 11 (4) (2002) 21–22.

[4] M.F. Chen, G.H. Tzeng, C.G. Ding, Combining fuzzy AHP with MDS in identifyingthe preference similarity of alternatives, Applied Soft Computing Journal 8 (1)(2008) 110–117.

[5] S.J. Chen, C.L. Hwang, F.P. Hwang, Fuzzy Multiple Attribute Decision makingMethods and Applications, Springer, New York, 1992.

[6] C.H. Cheng, Y. Lin, Evaluating the best main battle tank using fuzzy deci-sion theory with linguistic criteria evaluation, European Journal of OperationalResearch 142 (1) (2002) 174–186.

[7] T.H. Davenport, S.L. Jarvenpaa, M.C. Beers, Improving knowledge work pro-cesses, Sloan Management Review 37 (4) (1996) 53–65.

[8] M. Demarest, Understanding knowledge management, Long Range Planning30 (3) (1997) 374–384.

[9] K.C. Desouza, Strategic contributions of game rooms to knowledge manage-ment: some preliminary insights, Information & Management 41 (1) (2003)63–74.

10] A.C. Edmondson, A.B. Winslow, R.M.J. Bohmer, G.P. Pisano, Learning howand learning what: effects of tacit and codified knowledge on performanceimprovement following technology adoption, Decision Sciences 34 (2) (2003)197–224.

11] E. Fontela, A. Gabus, The DEMATEL Observer, DEMATEL 1976 Report. BATTELLEInstitute, Geneva Research Center, Geneva, Switzerland, (1976).

12] A. Gabus, E. Fontela, World Problems, An Invitation to Further Thought Withinthe Framework of DEMATEL. BATTELLE Institute, Geneva Research Centre,Geneva, Switzerland, (1972).

13] A. Gabus, E. Fontela, Perceptions of the World Problematique: Communica-tion Procedure, Communicating With Those Bearing Collective Responsibility(DEMATEL Report No. 1). BATTELLE Institute, Geneva Research Centre, Geneva,Switzerland, (1973).

14] C. Gopal, J. Gagnon, Knowledge, information, learning and the IS manager,Computerworld 29 (25) (1995) SS1–SS7.

15] K. Hafeez, H. Abdelmeguid, Dynamics of human resource and knowledge man-agement, Journal of the Operational Research Society 54 (2) (2003) 153–164.

16] P. Hess, J. Siciliano, Management: Responsibility for Performance, McGraw-Hill,New York, 1996.

17] C.W. Holsapple, K.D. Joshi, Knowledge manipulation activities: results of a Del-phi study, Information & Management 39 (6) (2002) 477–490.

18] S. Hori, Y. Shimizu, Designing methods of human interface for supervisorycontrol systems, Control Engineering Practice 7 (11) (1999) 1413–1419.

19] Y.C. Hu, R.S. Chen, G.H. Tzeng, Discovering fuzzy association rules using fuzzypartition methods, Knowledge-Based Systems 16 (3) (2003) 137–147.

20] C.Y. Huang, J.Z. Shyu, G.H. Tzeng, Reconfiguring the innovation policy portfoliosfor Taiwan’s SIP Mall industry, Technovation 27 (12) (2007) 744–765.

21] H.Y. Hu, Y.C. Lee, T.M. Yen, C.H. Tsai, Using BPNN and DEMATEL to modifyimportance-performance analysis model - A study of the computer industry,Expert Systems with Applications 36 (6) (2009) 9969–9979.

22] V. Hutchinson, P. Quintas, Do SMEs do knowledge management? Or simplymanage what they know? International Small Business Journal 26 (2) (2008)131–154.

23] B.D. Janz, P. Prasarnphanich, Understanding the antecedents of effective knowl-edge management: the importance of a knowledge-centered culture, Decision

Sciences 34 (2) (2003) 351–384.

24] J.A. Johannessen, J. Olaisen, B. Olsen, Mismanagement of tacit knowledge: theimportance of tacit knowledge, the danger of information technology, and whatto do about it, International Journal of Information Management 21 (1) (2001)3–20.

Compu

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

W.-W. Wu / Applied Soft

25] J.A Johannessen, B. Olsen, Knowledge management and sustainable compet-itive advantages: the impact of dynamic contextual training, InternationalJournal of Information Management 23 (4) (2003) 277–289.

26] J. Jordan, P. Jones, Assessing your company’s knowledge management style,Long Range Planning 30 (3) (1997) 392–398.

27] M. Kamaike, Design elements in the passenger car development: the classifica-tion and the influence analysis in case of recreational vehicle, Bulletin of JSSD48 (1) (2001) 29–38.

28] J.M. Kamara, C.J. Anumba, P.M. Carrillo, A CLEVER approach to selecting a knowl-edge management strategy, International Journal of Project Management 20 (3)(2002) 205–211.

29] C.W. Li, G.H. Tzeng, Identification of a threshold value for the DEMATEL methodusing the maximum mean de-entropy algorithm to find critical services pro-vided by a semiconductor intellectual property mall, Expert Systems withApplications 36 (6) (2009) 9891–9898.

30] R.J. Li, Fuzzy Method in group decision making, Computers and Mathematicswith Applications 38 (1) (1999) 91–101.

31] S.H. Liao, Problem solving and knowledge inertia, Expert Systems with Appli-cations 22 (1) (2002) 21–31.

32] S.H. Liao, Knowledge management technologies and applications-literaturereview from 1995 to 2002, Expert Systems with Applications 25 (2) (2003)155–164.

33] C. Lin, B. Tan, P.J. Hsieh, Application of the fuzzy weighted average in strategicportfolio management, Decision Sciences 36 (3) (2005) 489–511.

34] J.J.H. Liou, L. Yen, G.H. Tzeng, Building an effective safety management systemfor airlines, Journal of Air Transport Management 14 (1) (2008) 20–26.

35] C.L. Lin, G.H. Tzeng, A value-created system of science (technology) park byusing DEMATEL, Expert Systems with Applications 36 (6) (2008) 9683–9697.

36] C.J. Lin, W.W. Wu, A causal analytical method for group decision-making underfuzzy environment, Expert Systems with Applications 34 (1) (2008) 205–213.

37] R. Maier, U. Remus, Implementing process-oriented knowledge managementstrategies, Journal of Knowledge Management 7 (4) (2003) 62–74.

38] M. Martensson, A critical review of knowledge management as a managementtool, Journal of Knowledge Management 4 (3) (2000) 204–216.

39] I. Nonaka, A dynamic theory of organizational knowledge creation, Organiza-tion Science 5 (1) (1994) 14–37.

40] I. Nonaka, N. Konno, The concept of “ba”: building a foundation for knowledgecreation, California Management Review 40 (3) (1998) 40–54.

41] I. Nonaka, R. Toyama, N. Konno, SECI, Ba and leadership: a unified model ofdynamic knowledge creation, Long Range Planning 33 (1) (2000) 5–34.

42] S. Opricovic, G.H. Tzeng, Defuzzification within a multicriteria decision model,International Journal of Uncertainty Fuzziness and Knowledge-Based Systems11 (5) (2003) 635–652.

43] S. Opricovic, G.H. Tzeng, Compromise solution by MCDM methods: a compar-ative analysis of VIKOR and TOPSIS, European Journal of Operational Research156 (2) (2004) 445–455.

44] M. Oussalah, On the compatibility between defuzzification and fuzzy arith-metic operations, Fuzzy Sets and Systems 128 (2) (2002) 247–260.

45] People Management, The people factor, People Management 4 (2) (1998) 38.46] S. Raub, D. Von Wittich, Implementing knowledge management: three strate-

gies for effective CKOs, European Management Journal 22 (6) (2004) 714–724.47] R. Roos, J. Roos, Measuring your company’s intellectual performance, Longe

Range Planning 30 (3) (1997) 413–426.48] B. Rubenstein-Montano, J. Liebowitz, J. Buchwalter, D. McCaw, B. Newman, K.

Rebeck, The Knowledge Management Methodology Team, A systems think-ing framework for knowledge management, Decision Support Systems 31 (1)(2001) 5–16.

[[

ting 12 (2012) 527–535 535

49] R. Sabherwal, I. Becerra-Fernandez, An empirical study of the effect of knowl-edge management processes at individual, group, and organizational levels,Decision Sciences 34 (2) (2003) 225–260.

50] S.M Seyed-Hosseini, N. Safaei, M.J. Asgharpour, Reprioritization of failures in asystem failure mode and effects analysis by decision making trial and evalua-tion laboratory technique, Reliability Engineering & System Safety 91 (8) (2006)872–881.

51] I. Spiegler, Technology and knowledge: bridging a “generating” gap, Informa-tion & Management 40 (6) (2003) 533–539.

52] D.J. Teece, Strategies for managing knowledge assets: the role of firm structureand industrial context, Long Range Planning 33 (1) (2000) 35–54.

53] W.H. Tsai, W.C. Chou, Selecting management systems for sustainable develop-ment in SMEs: a novel hybrid model based on DEMATEL, ANP, and ZOGP, ExpertSystems with Applications 36 (2) (2009) 1444–1458.

54] W.H Tsai, J.L. Hsu, Corporate social responsibility programs choice and costsassessment in the airline industry-A hybrid model, Journal of Air TransportManagement 14 (4) (2008) 188–196.

55] M.L. Tseng, Using a hybrid MCDM model to evaluate firm environmentalknowledge management in uncertainty, Applied Soft Computing 11 (1) (2011)1340–1352.

56] M.L. Tseng, Implementation and performance evaluation using fuzzy networkbalanced scorecard, Computers & Education 55 (1) (2010) 188–201.

57] M.L. Tseng, Using linguistic preferences and grey relational analysis to evaluatethe environmental knowledge management capacities, Expert Systems withApplications 37 (1) (2010) 70–81.

58] M.L. Tseng, An assessment of cause and effect decision making model for firmenvironmental knowledge management capacities in uncertainty, Environ-mental Monitoring and Assessment 161 (1–4) (2010) 549–564.

59] M.L. Tseng, Application of ANP and DEMATEL to evaluate the decision-makingof municipal solid waste management in Metro Manila, Environmental Moni-toring and Assessment 156 (1–4) (2009) 181–197.

60] M.L. Tseng, Y.H. Lin, Application of Fuzzy DEMATEL to develop a cause and effectmodel of municipal solid waste management in Metro Manila, EnvironmentalMonitoring and Assessment 158 (1–4) (2009) 519–533.

61] C.V. Von Altrock, Fuzzy Set Theory & Neurofuzzy Applications in Business andFinance, Prentice Hall, New Jersey, 1996.

62] K.M. Wiig, Integrating intellectual capital and knowledge management, LongRange Planning 30 (3) (1997) 399–405.

63] K.M. Wiig, R.D. Hoog, R.V.D. Spek, Supporting knowledge management: a selec-tion of methods and techniques, Expert Systems With Applications 13 (1)(1997) 15–27.

64] W.W. Wu, Choosing knowledge management strategies by using a combinedANP and DEMATEL approach, Expert Systems with Applications 35 (3) (2008)828–835.

65] W.W. Wu, Y.T. Lee, Developing global managers’ competencies using thefuzzy DEMATEL method, Expert Systems with Applications 32 (2) (2007)499–507.

66] M. Yamazaki, K. Ishibe, S. Yamashita, I. Miyamoto, M. Kurihara, H. Shindo,An analysis of obstructive factors to welfare service using DEMATEL methodreports of the faculty of engineering, Yamanashi University 48 (1997)25–30.

67] A. Yuzawa, A state and subjects of TMO conception for city core vitalization

countermeasure: a case study of Maebashi TMO conception, Bulletin of Mae-bashi Institute of Technology 5 (2002) 61–67.

68] L.A. Zadeh, Fuzzy Sets, Information and Control 8 (2) (1965) 338–353.69] L.A. Zadeh, The concept of a linguistic variable and its application to approxi-

mate reasoning, Information Science 8 (1975) 199–249, 301–357 (II).