A fuzzy trust evaluation method for knowledge sharing in virtual enterprises

Computers & Industrial Engineering 59 (2010) 853–864

Contents lists available at ScienceDirect

Computers & Industrial Engineering journal homepage: www.elsevier.com/locate/caie

A fuzzy trust evaluation method for knowledge sharing in virtual enterprises Tsung-Yi Chen a,⇑, Yuh-Min Chen b, Chia-Jou Lin b, Pin-Yuan Chen b a b

Department of Electronic Commerce Management, Nanhua University, Chia-Yi, Taiwan, ROC Institute of Manufacturing Information and Systems, National Cheng Kung University, Tainan, Taiwan, ROC

a r t i c l e

i n f o

Article history: Received 10 February 2010 Received in revised form 15 July 2010 Accepted 26 August 2010 Available online 8 September 2010 Keywords: Virtual enterprise Knowledge sharing Trust Fuzzy theory

a b s t r a c t The success of virtual enterprises (VEs) depends on the effective sharing of related resources between various enterprises or workers who perform related activities. Specifically, VE success hinges on the integration and sharing of information and knowledge. Trust is an important facilitator of knowledge sharing. However, the trustworthiness of a peer is a vague concept that is dynamic and that often shifts over time or with environmental changes. This study designs a trust-based knowledge-sharing model based on characteristics of VEs and the knowledge structure model to express knowledge associated with VE activities. Subsequently, the factors that affect the trust evaluation are identified based on the characteristics of trust and VEs. Finally, this study develops a knowledge sharing, decision-making framework in which a fuzzy trust evaluation method for sharing knowledge is proposed based on VE activities and the interactions among workers in allied enterprises. The method consists of three sub-methods, including an activity correlation evaluation method, a current trust evaluation method, and an integral trust evaluation method. Under the premises of secure VE knowledge and reasonable access authorization, the proposed knowledge-sharing method provides the trust level between a knowledge-requesting enterprise and a knowledge-supplying enterprise to improve the willingness of the latter to share more valuable knowledge, ultimately increasing the efficiency and competitiveness of VEs. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Virtual enterprises (VEs) represent a dynamic networking alliance that can react sensitively to changing market opportunities and gather knowledge resources from a wide range of enterprises using Internet technology to develop, design, manufacture, and market goods and services (Yang & Lin, 2008). The success of a VE depends on the effective sharing of related resources between activities performed by various enterprises and, in particular, on the integration and sharing of information and knowledge among alliance enterprises. Knowledge sharing refers to the exchange and discussion of knowledge among members of an organization, between internal and external teams, or between organizations for the purpose of improving organizational competiveness by the effective exchange, integration, and synergy of knowledge (Chen, 2008; Lawson, Petersen, Cousins, & Handfield, 2009). Knowledge sharing is difficult to implement. Previous investigations of enterprise knowledge sharing have tended to focus on the deployment of

⇑ Corresponding author. Address: Department of Electronic Commerce Management, Nanhua University, No. 55, Sec. 1, Nanhua Rd., Zhongkeng, Dalin Township, Chiayi County 62248, Taiwan, ROC. Tel.: +886 5 2721001x56440; fax: +886 5 2427197. E-mail address: [email protected] (T.-Y. Chen). 0360-8352/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.cie.2010.08.015

information technological infrastructure, such as document management systems, information search technologies, and forums, to improve the environment for knowledge management (Fla’via Maria, Marcos, Borges, & Erick, 2006; Gollmann, 1999; Lin, Wang, & Tserng, 2006). Strader, Lin and Shaw (1998) and Chen (2008) adopted an access control perspective to investigate knowledge access authorization of users to assist enterprises in knowledge sharing. Knowledge sharing within VEs is determined by key factors such as the VE process and trust among the enterprises. Trust has been defined as a psychological state that comprises the intention to accept vulnerability based on positive expectations regarding the intentions or behavior of others without the ability to monitor or control that other party (Zolin, Hinds, Fruchter, & Levitt, 2004). Knowledge sharing in distributed environments requires more a priori trust than face-to-face communication (Riegelsberger, Sasse, & McCarthy, 2003). Investigations have found that a higher level of trust corresponds to greater willingness to share knowledge (Cheng, Hailin, & Hongming, 2008; Quigley, Tesluk, Locke, & Bartol, 2007; Willem & Buelens, 2007). However, trust depends on an implicit set of beliefs, which are vague. Trust is a multi-dimensional construct (Mayer, Davis, & Schoorman, 1995; Kanawattanachai & Yoo, 2002) and has various definitions that are appropriate to different application domains (Ford, 2003). Trust varies with time, the environment, and other factors. Therefore,

854

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

objectively evaluating trust has become an important issue in the field of knowledge sharing. To integrate knowledge that is distributed across allied enterprises while providing the VE members with authorized access to knowledge that is related to or required for tasks of interest, this study proposes a knowledge-sharing method for VEs that is based on the process of VEs and considers trust among allied enterprises. To develop the methodology, a knowledge-sharing model is designed based on the characteristics of a VE, and then a structure of activity-related knowledge is designed to express the knowledge of VE activities. Subsequently, the factors that affect the trust evaluation are identified, derived from the characteristics of trust and VEs. Because trust always involves ambiguity and subjectivity and is difficult to estimate experimentally by modeling some graded phenomenon, trust degree cannot be specifically measured using crisp values. Therefore, a fuzzy-based trust evaluation method is developed; it includes a method for evaluating the correlation among activities, the current trust evaluation method, and the integral trust evaluation method. Assuming secure VE knowledge and reasonable access authorization, the proposed knowledgesharing approach provides a decision-making support framework centered on trust evaluation between a knowledge requester and a knowledge supplier that will improve the willingness of the latter to sharing valuable knowledge and consequently will increase the efficiency and competitiveness of VEs. 2. Related literature This section surveys a number of related studies of VE activity and knowledge, trust, knowledge sharing, and fuzzy theory. 2.1. Activity and knowledge in VEs Activity theory is a set of basic principles for constituting a general conceptual system (Kaptelinin & Nardi, 1997, 2006; Holland & Reeves, 1994). Beckett (2004) used the theory as a framework for discussion of the organizational attributes associated with VE operation. The unit of analysis in activity theory is an activity directed at an object which motivates activity. An activity contains various artifacts, for example instruments, signs, procedures, machines, methods, laws, forms of work organization (Nardi, 1996). Activities are composed of goal-directed actions. Different actions may be performed by different VE workers to meet the same goal. In activity theory, the constituents of activity are not fixed, but can dynamically changes with conditions (Kaptelinin & Nardi, 1997). In this study, activity theory is helpful to understand and analyze different kinds of VE activities. Activities of VEs must be analyzed to provide an understanding of the knowledge that is required for particular activities. Knowledge can be structured experiences, values, text-based information, or unique expert insights. It resides in not only documents that are stored in a knowledge management system but also in daily routine tasks, processes, executions, and norms (Davenport & Prusak, 1998; Lee, 2001). Since categories of knowledge vary with perspective, this study considers three dimensions in categorizing knowledge in a VE. (1) Abstractness. This dimension can be divided into (a) formal knowledge: conceptual knowledge that is derived by the generalization, analysis, and validation of data collected by scientifically objective means and (b) practical knowledge: specific job skills, experience-based rules, causal relationships, or input/output of enterprise activities derived from practices and generally preserved in knowledge cases and personal experiential knowledge databases (Beckman, 1997).

(2) Phenomenon comprehension and application purpose, which is divided into (a) declarative knowledge (Know-what): concepts, composition, and structure of an event; (b) causal knowledge (Know-why): knowledge of causes and consequences of an event; (c) procedural knowledge (Know-how): knowledge of processes, steps, and methods associated with the execution of an event; and (d) relational knowledge (Know-with): knowledge of relationships between an event and other important factors (Quinn, Anderson, & Finkelstein, 1996). (3) Openness: Given the need for some knowledge to be securely protected, knowledge can be divided into (a) public knowledge: defined as knowledge related to the VE project that all member enterprises must provide and share and (b) private knowledge: techniques or knowledge related to the VE project that are owned but not directly shared by enterprises, which can thus decide whether or not to share such knowledge based on an evaluation of trust with another party. 2.2. Trust Trust has been defined in various ways for various situations and specific contexts. Trust is a multi-dimensional and multi-level dynamic concept (Lewicki & Bunker, 1996; Butler, 1991). Mayer et al. (1995) claimed that trust comprises ability, benevolence, and integrity. Mishra (1996) extended this concept by defining four dimensions of trust—concern, reliability, competence, and openness. Meyerson, Weick, and Kramer (1996) proposed the concept of swift trust, which applied to members of temporary teams, who tend to relate to each other according to roles rather than as individuals. Accordingly, a specific definition of trust pertaining to employees in an organization involves positive expectations, such as integrity, capability, truthfulness, goodwill, and ability, which relate to the competence and reliability of fellow employees within the organization (Ellonen, Blomqvist, & Puumalainen, 2008). Chowdhury (2005) identified two main forms of trust: (1) cognitive trust, based on cognitive reasoning regarding reliability of performance and competence and (2) affective trust, based on emotional ties with someone. Koehn (2003) investigated four forms of trust: (1) goal-based trust, which appears between two people who think they share a common objective; (2) calculative trust, which attempts to predict what the trusted party will do by seeking evidence of the other’s trustworthiness; (3) knowledge-based trust, which arises when people are familiar with each other and/or interact frequently; and (4) respect-based trust, which is reinforced when the two parties in a relationship have a similar love of virtue, excellence, and wisdom and are willing to engage in dialogue and ongoing conversation with a view to understanding each other better. 2.3. Trust associated with knowledge sharing Trust evaluation is a valuable means of promoting knowledge sharing (Gruber, 2000; Ling, San, & Hock, 2009; McEvily, Perrone, & Zaheer, 2003). Renzl (2008) provided empirical evidence that trust in management facilitates knowledge sharing by reducing fear of loss of one’s unique value. Restated, a trusting person is more willing to provide useful knowledge to others. Newell, David, and Chand (2007) and Lin (2008) investigated issues related to trust and the sharing of knowledge in globally distributed IT work teams and developed a threefold typology of trust that included commitment, companion, and competence trust. Commitment trust is based on contractual agreements between members who expect to derive mutual benefits from their cooperative relationship. Commitment trust can reduce team risk and uncertainty

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

because contractual agreements can restrict the behavior of individual team members. However, trust can be dynamic, changing with time (Boon & Holmes, 1991; Chang, Hussain, & Dillon, 2005; Lewicki & Bunker, 1996). Robust trust is established over time, and collaborative experiences support the sharing of more sensitive knowledge (Dodgson, 1993). However, trust may decrease if it was established a long time ago. Based on the above studies, trust is an important facilitator of knowledge sharing (Bakker, Leenders, Gabbay, Kratzer, & Engelen, 2006; Lin & Tseng, 2005). However, trust between allied enterprises and trust between employees are equally important to knowledge sharing in VEs. 2.4. Fuzzy theory for evaluating trust Various factors can influence the degree of interpersonal trust, including cultural diversity, interaction experiences, cooperation and communication, worker propensity to trust, worker perception of partner trustworthiness, and risk and reward (Mayer et al., 1995; Zolin et al., 2004). The complexity of trust-related beliefs is responsible for the ‘‘fuzziness” of trust: trust cannot be specifically measured using crisp values, because it always involves ambiguity and subjectivity and is difficult to estimate experimentally by modeling some graded phenomenon. To solve this problem, fuzzy theory (Zadeh, 1965) has been utilized to evaluate trust in various fields (Flaminio, Pinna, & Tiezzi, 2008; Luo, Liu, & Fan, 2009; Victor, Cornelis, Cock, & Silva, 2009). Fuzzy logic is a highly intuitive approach to analysis using natural language labels that represent intervals rather than exact values (Castelfranchi, Falcone, & Pezzulo, 2003). To promote VE efficiency, Mun, Shin, Lee, and Jung (2009) proposed a fuzzy method for evaluating trust to support collaboration among enterprises and to maximize the satisfaction of cooperation among members. The method is useful in selecting good partners. In e-commerce, Bharadwaj and Al-Shamri (2009) presented fuzzy computational models of both trust and reputation to enhance the security of online transactions. Schmidt, Steele, Dillon, and Chang (2007) developed a model for calculating the trustworthiness of potential business partners using fuzzy concepts to facilitate the selection of the best-matched and most trustworthy business partner. Trust changes with the passage of time; hence, trust is both fuzzy and dynamic (Chang et al., 2005). Wei, Lu, and Yanchun (2008) developed an approach based on a fuzzy cognitive time maps (FCTMs) theory that accounted for the dynamic nature of trust to analyze the evolution of trust in VEs. 3. Analysis and design of the knowledge-sharing model in VEs In the section, activity theory (Kaptelinin & Nardi, 1997, 2006) is applied to support the analysis of knowledge-sharing model relative to VE activity, this section firstly elucidates the design of a knowledge-sharing model that describes the relationships among activities, roles, and knowledge in a VE. It then presents the structure of knowledge associated with activities. Lastly, it identifies trust factors and presents the knowledge sharing, decision-making support framework for evaluating trust among alliance enterprises that share knowledge. 3.1. Knowledge-sharing model in VEs Knowledge sharing is the activity of transferring or disseminating knowledge from one person, group, or organization to another (Lee, 2001). In VE operations, knowledge is required for executing activities. Knowledge associated with activities may exist as (1) employees’ knowledge, including skills, experiences, habits and in-

855

stincts, or (2) organizational knowledge, including the intellectual property of the enterprise, which is often stored in an enterprise knowledge base in document or digital format. This study focuses on the sharing of organizational knowledge. Channels of knowledge sharing can be divided into (1) central control knowledge replication, in which knowledge of standard contents is transmitted and shared within an organization through formal channels, passing on important knowledge to its members, and (2) autonomous knowledge networks, in which the members of an organization directly share knowledge (Probst, Raub, & Romhardt, 2000). This study focuses mainly on a knowledge-sharing model that features central control knowledge replication based on the characteristics of VEs. Upon the formation of a VE, a worker in an alliance of various enterprises is assigned a VE role. The worker is responsible for executing the associated activities, which require the exploitation of certain resources and knowledge to achieve the joint objectives of the VEs. Knowledge must be shared at the right time, at the right place, and among the right people. Whether knowledge should be shared at all depends on the relationships among roles, activities, and knowledge within a VE. Based on the above analyses, this study designs a knowledgesharing model (Fig. 1) that includes roles, activities, and knowledge within a VE, according to which relevant knowledge is provided to support VE workers in executing tasks. Once members of enterprises have been assigned VE roles in which to execute VE activities collaboratively, they can share activity-related knowledge to ensure the smooth completion of their tasks. For example, when a member of enterprise A is assigned VE role1 and is responsible for executing VE activity 1, knowledge related to VE activity 1 must be shared with the member. Relationships among VE activities include sequence, branch, and join relationships. Relationships among activities are associated with authorizations to share knowledge. For example, if VE activity 1 and VE activity 2 are correlated sequentially, then the worker who performs VE activity 1 can share knowledge associated with VE activity 2. Knowledge sharing should break down the boundaries between allied enterprises in VEs, because workers from different enterprises can be assigned the same VE role, and the same VE activity can be jointly executed by different VE roles. However, enterprises store activity-related knowledge, including both public and private knowledge. When private knowledge must be shared, the trust between two enterprises should be evaluated to support a knowledge-sharing decision. For example, the private knowledge associated with VE activity 1 is stored in enterprise A. When the worker with VE role 2 (VER 2) wants to share this private knowledge, he/she musts ask enterprise A, which then evaluates its trust with enterprise B to determine whether the private knowledge can be shared. This study focuses mainly on the sharing of private knowledge among enterprises. 3.2. Activity-related knowledge structure To effectively store, organize, manage, and use the knowledge on which various activities depend, this study constructs an activity-related knowledge structure (Fig. 2) using ontological technology to express various categories of knowledge in VEs and relationships between activities and knowledge, based on the three dimensions of VE knowledge that were described in Section 2.1. In the activity-related knowledge structure, each knowledge category contains an ID, properties and instances. Each piece of activityrelated knowledge contains formal knowledge and practical knowledge. Formal knowledge refers to product knowledge and principle knowledge associated with activities (Davenport & Prusak, 1998; Lee, 2001). Product knowledge includes descriptions of a product,

856

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

Enterprise B Enterprise A e ask/r

ply

Enterprise C Knowledge

VER 3

VER 2

VER 1

VER 7

Knowledge

Private knowledge

Knowledge Activity 3

Activity 2

Public knowledge

Activity 1

Activity 6

Activity 7 End

Activity 4 Activity knowledge

Branch

Sequence

Activity 5 Activity knowledge

Assignment Authorization

Join

Belong to

Is-a

Part-of

Execution

Fig. 1. Knowledge-sharing model in VEs.

Activity Knowledge Part-of

Part-of ID

Formal Knowledge

Product Knowledge

Declarative Knowledge (Know-What)

Principle Knowledge

Causal Knowledge (Know-Why)

Property

Practical Knowledge

Instance

Empirical Knowledge

Case

Procedural Knowledge (Know-How)

Relational Knowledge (Know-With)

Fig. 2. Activity-related knowledge structure.

its features and functions, its R&D, its realization, and its use and maintenance. Principle knowledge includes standard operating procedures and engineering principles. Practical knowledge includes cases and empirical knowledge (Beckman, 1997). A case is a formal record of the organized

execution processes of a previous project, and it serves as a reference for executing a similar task or answering related questions. A case includes both the task description and the solution. In terms of the purpose of using knowledge, the task description of a particular case includes Know-what and Know-why, while the solution

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

includes the Know-how and its corresponding Know-why (Quinn et al., 1996). In this study, each case is divided into four parts: (1) Know-what: basic description of the case; (2) Know-why: inputs and outputs associated with the case; (3) Know-how: process by which the case is solved; and (4) Know-with: theoretical knowledge that is related to the case. Empirical knowledge refers to various experiences and personal understandings. In this study, empirical knowledge is defined as knowledge that is stored in work logs and reports of actual job executions by individuals. Since this knowledge is neither organized nor generalized, this study only refers to informal practical knowledge. 3.3. Trust-based decision-making support framework for knowledge sharing This section identifies factors that influence the degree of trust based on the proposed knowledge-sharing model and knowledge structure. A knowledge-sharing decision procedure is then designed based on the proposed factors. 3.3.1. Identification of trust factors Factors that are adopted to evaluate trust vary with the application domain. Trust in VEs is very complex and cannot be easily evaluated because of its inter-organizational and distribution characteristics (Au, Looi, & Ashley, 2001). The six characteristics of trust are implicitness, asymmetry, transitivity, antonymy, asynchrony, and gravity. Trust is fuzzy and dynamic (Chang et al., 2005). Therefore, in accordance with the proposed knowledge-sharing model and characteristics of trust, this study identifies trust factors, as follows: (1) Direct relationship between enterprises In VEs, trust is affected directly by the relationships between two enterprises and the relationships between activities that are performed by the two enterprises. Two factors are defined in terms of a direct relationship between two enterprises, as follows:  Collaborative relation (CR): This relation refers to the way in which one enterprise works with another. It may include outsourcing, collaboration, and corporate relations. The outsourcing relation may exist when two enterprises are involved in different processes in VEs. The collaboration relation refers to the fact that two enterprises may work on different activities within a single process in VEs. The corporate relation appears when two enterprises work on the same activity in a single process in VEs. CR determines the degree of interaction between the two enterprises. Different relations between enterprises are associated with different levels of trust.  Activity correlation (AC): Greater correlation among activities in a process is associated with a stronger interactive relationship between enterprises that are responsible for executing these activities. Greater interactive relationship between enterprises is associated with greater trust. (2) Indirect relationship between enterprises If one enterprise has no collaborative experience with a particular firm, then the trust between these two enterprises can be evaluated by the collaborations that the firm has with other enterprises. Therefore, others’ trust value (OTV) is one factor in evaluating trust for knowledge sharing in VEs. (3) Dynamic relationship between enterprises Trust is not constant and often changes over time. Therefore, this study defines two factors that affect trust—collaborative time and past trust value.

857

 Collaborative time (CT): Trust is accumulated during time spent in collaboration. A longer period of collaboration corresponds to greater trust between two enterprises.  Past trust value (PTV): Trust between enterprises is affected by previous experiences of collaboration. Therefore, PTV is adopted to evaluate trust between enterprises. 3.3.2. Knowledge-sharing decision-making framework Using the proposed trust factors, this study designs a knowledge-sharing decision-making framework for VEs (Fig. 3) that comprises mainly the activity correlation evaluation module, the current trust evaluation module, the trust level assessment module, and two databases: the VE Base and the Physical Knowledge Base. The VE base records VE-related data, including VE objectives, activities, roles, task assignments, and schedules. The Physical Knowledge Base contains private knowledge of individuals. The activity correlation evaluation module performs the activity relation calculation, the instance similarity calculation in activity-related knowledge, the property similarity calculation in activityrelated knowledge, and the activity comparison, which are all used in the evaluation of the strength of the relationship between activities that are executed by the knowledge requester and by the knowledge owner. The current trust evaluation module then calculates the trust between two enterprises based on the activity correlation, the collaboration relation, and the collaboration time. Finally, to evaluate the objective trust value, the trust level assessment module adjusts the current trust value according to past trust values between two enterprises and the trust values provided by other enterprises. The trust level is used in a decision about sharing knowledge. Each module performs one task, as detailed in the next section. 4. Methods for evaluating relationships between activities The purpose of this method is to evaluate the relationship between two activities that are executed by a knowledge requester and a knowledge owner. This method consists of four steps, described below. 4.1. Calculation of activity relation (Step 1) All activities associated with a process in VEs can be mutually related in a network. Activities are represented by vertices and relationships between two activities are represented by edges that link a pair of two vertices. The strength of the relationship between two activities is determined by the distance and type of relationship between the two activities. A longer edge corresponds to a stronger correlation between the two activities. The set E = {er:er 2 R and 0 6 er 6 1} is defined as the length of the edge between two activities, where r represents a sequential, branched, or joint relationship. The lengths (er) are determined by the VE administrator in the VE formation stage, according to his/her perception of the activities. Since multiple paths may exist between two vertices, Dijkstra’s algorithm (Agnarsson & Greenlaw, 2007) is utilized to calculate the shortest path between two activities as the distance between them. The starting vertex in this study is the activity executed by the knowledge requester, and the ending vertex is the activity associated with the desired knowledge. Dijkstra’s algorithm calculates the shortest path from the starting vertex to the ending vertex, which is given by (F.1).

dðx; SÞ ¼ minfdðx; uÞ þ wðu; v Þg; u2S

v 2S

ðF:1Þ

858

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

Activity Correlation Evaluation Module

Current Trust Evaluation Module

Trust Level Assessment Module

Activity Relation Calculation VE Role Activity Knowledge Instance Similarity Calculation

Activity Correlation Calculation

Activity Knowledge Property Similarity Calculation

VE Base

Collaborative Relation Identification

Trust Calculation

Collaborative Time Evaluation

Past Trust Value Calculation

Trust Adjustment

Other Trust Value Calculation

Physical Knowledge Base

VE Activity Knowledge Structure

Enterprises

Fig. 3. Knowledge-sharing decision-making framework.

where dðx; SÞ is the shortest path from starting node x to node S in a network (V); V is the set of nodes in a network; S is all nodes in a network V, excluding node S; and w(u, v) is the shortest distance from node u to node v. After the shortest distance between two activities has been determined, the activity relation function (F.2) can be adopted to calculate the strength of the relationship between them.

ar pq ¼

4.4. Calculation of correlation between activities (Step 4)

1 ; 1 þ adpq

ðF:2Þ

where a is a weighting factor that is determined by the VE leader and dpq is the shortest distance from activity p to activity q. 4.2. Calculation of similarity between instances of activity knowledge (Step 2) To calculate the relationship between the knowledge associated with two activities, a VE activity knowledge similarity algorithm is proposed based on Jaccard Coefficient (Guha, Rastogi, & Shim, 1998). The similarity between the knowledge associated with two activities is calculated using the two algorithms for property similarity and instance similarity. First, the correlation between the knowledge associated with two activities is calculated using (F.3).

InstanceSmðIp ; Iq Þ ¼

where Pp is the set of properties of knowledge associated with activity p; for example, Pp = {pp1, pp2, . . . , ppm}, where ppi represents the ith property of knowledge associated with activity p and Pq is the set of properties of knowledge associated with activity q; for example, Pq = {pq1, pq2, . . . , pqn}, where pqj represents the jth property of knowledge associated with activity q.

jIp \ Iq j ; jIp [ Iq j

for p – q;

ðF:3Þ

where Ip is the set of instances of knowledge associated with activity p; for example, Ip = {ip1, ip2, . . . , ipm}, where ipj represents the jth instance of knowledge associated with activity p; and Iq is the set of instances of knowledge associated with activity q; for example, Iq = {iq1, iq2, . . . , iqn}, where iqk represents the kth instance of knowledge associated with activity q. 4.3. Calculation of similarity between properties of activity knowledge (Step 3) The correlation between the knowledge associated with two activities is calculated using (F.4).

PropertySmðPp ; Pq Þ ¼

jPp \ Pq j ; jPp [ Pq j

for p – q;

ðF:4Þ

The activity correlation calculation (F.5) involves the activity relation, the similarity between instances of knowledge associated with activities, and the similarity between properties of knowledge associated with activities, obtained using Functions 2–4 to determine the strength of the correlation between activities. Function 5 must be constrained by Eq. (6), whose three weighting factors are determined by the VE leader.

ACpq ¼ wr arpq þ wi InstanceSmðIp ; Iq Þ þ wp PropertySmðP p ; P q Þ; wr þ wi þ wp ¼ 1;

ðF:5Þ ðE:6Þ

where wr, wi, and wp are weighting factors of the activity relation, the similarity between instances of knowledge associated with activities, and the similarity between properties of knowledge associated with activities, respectively. 5. Method for evaluating current trust This approach evaluates the current trust between the party that requires knowledge and the owner of that knowledge. The current trust calculation is developed herein using fuzzy theory. The approach is divided into two steps, described below. 5.1. Determination of fuzzy sets (Step 1) First, CR, CT, and AC are defined as input variables, and CTR is defined as the output variable. The fuzzy set and membership functions are as follows: (1) CR: The fuzzy set and memberships functions in CR are defined below:

859

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

e ¼ CR

3 X

lcrj

j¼1

cr j

;

for cr j 2 U cr ; 0 6 lcr j 6 1; j ¼ 1; 2; 3;

ðF:7Þ

In this example, the membership function of CTR is defined as shown in Fig. 4c.

e R is the fuzzy set of collaborative relations; crj is the where C collaborative relation—cr1, cr2, and cr3 represent outsourcing (o), collaboration (col), and corporate (cor), respectively; lcr j is the membership value of the collaborative relation; and Ucr is the universe of the collaborative relation; in this example, Uco = {o, col, cor}. In this example, which includes three CRs, the fuzzy set of the collaborative relation is assumed to be

5.2. Fuzzy inference (Step 2) Fuzzy inference is the calculation of fuzzy relations based on logical rules in a fuzzy rule bank. Mamdani-style Inference (Negnevitsky, 2005) is adopted in this study because this method can be directly applied to logical rules made by human beings, rendering the leverage of existing expert knowledge easy. The major steps of Mamdani-style Inference are as follows.

e ¼ 0:15 þ 0:5 þ 0:85 : CR o col cor

(1) Fuzzification: the corresponding semantic fuzziness and membership values of the three input variables CR, CT, and AC are derived from membership functions defined at Step 1. (2) Rule evaluation: each rule consists of two parts—the antecedent and the consequent:

As indicated by the above equation, the corporate relation yields the highest trust with a membership value of 0.85, while the outsourcing relation yields the lowest trust with a membership value of 0.15. (2) CT: The linguistic variable associated with the fuzzy set of collaborative time is T(ct) = {short, medium, long}. The membership function is defined in terms of triangular fuzzy numbers, as follows:

IF < antecedent >; THEN < consequent > Fuzzified inputs are used in the antecedents of the fuzzy rules. As an example, Table 1 shows the fuzzy rules for inferring the current trust between two enterprises based on expert opinions. These rules are stored in a fuzzy rule base. (3) Aggregation of rule outputs: the fitness values of all rules are obtained from the membership function and summarized in a useful reference. As shown in (E.11) and (E.12), the first step is to take the logical product (min operation) and calculate the antecedent, or membership value Mi, and the consequent, or membership value Ni, of each fuzzy rule.

lCT TðctÞ ¼ faTðctÞ ; bTðctÞ ; c TðctÞ g; where aTðctÞ < bTðctÞ < cTðctÞ :

ðF:8Þ

In this example, the unit of collaborative time is a year. This study assumes that if the period of collaboration of two enterprises exceeds ten years, then these two enterprises have collaborated for a long time. Therefore, the membership function of CT is defined as shown in Fig. 4a. (3) AC: The linguistic variables that are associated with the fuzzy set of the activity correlation is T(ac) = {weak, moderate, strong}. The membership function is defined in terms of triangular fuzzy numbers, as follows:



CT AC M i ¼ lCR T ðcrÞ ^ lT ðctÞ ^ lT ðacÞ ¼ min

ðE:11Þ

ðF:9Þ

In this example, the membership function of AC is defined as shown in Fig. 4b. (4) CTR: The linguistic variables that are associated with the fuzzy set of current trust is T(ctr) = {low, medium, high}. The membership function is defined in terms of triangular fuzzy number as follows:

U i ¼ maxfNi g;

for i ¼ 1; 2; 3; . . . ; n:

Pn



COG ¼ CTRV ¼

ðF:10Þ

µ Short

1

0

Medium

5

(a)

Weak

Long

10

year

0

ðE:13Þ

(5) Defuzzification: the output of the aggregated rules is defuzzificated using the Center of Gravity (COG) method (E.14) that is the most defuzzification method; COG is the central value of a fuzzy set. The result is a specific output value (CTRV).

lCTR where aTðctrÞ < bTðctrÞ < cTðctrÞ : TðctÞ ¼ aTðctÞr ; bTðctÞr ; cTðctÞr ;

µ

ðE:12Þ

Then, the logical sum (max operation) of the consequents can be derived and expressed as a membership value (Ui) in (E.13).







CT AC lCR T ðcrÞ; lT ðctÞ; lT ðacÞ ;

N i ¼ minfM i ; lCTR T g:

lAC TðacÞ ¼ aTðacÞ ; bTðacÞ ; c TðacÞ ; where aTðacÞ < bTðacÞ < cTðacÞ :







lCTR T ðxj Þ : CTR j¼1 lT ðxj Þ

j¼1 xj Pn

ðE:14Þ

µ Moderate

0.5

Strong

1

(b) Fig. 4. Triangular fuzzy numbers.

Low

Medium

0.5

(c)

High

1

860

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

Table 1 Fuzzy rules for evaluating current trust.

Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule Rule

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

CR

CT

AC

CTR

O O O O O O O O O Col Col Col Col Col Col Col Col Col Cor Cor Cor Cor Cor Cor Cor Cor Cor

Short Short Short Medium Medium Medium Long Long Long Short Short Short Medium Medium Medium Long Long Long Short Short Short Medium Medium Medium Long Long Long

Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong Weak Moderate Strong

Low Low Low Low Low Low Low Low Low Low Medium Medium Medium Medium Medium Medium Medium High Medium High High High High High High High High

6. Method for evaluating integral trust In addition to current trust, trust between two enterprises is affected by past experience and opinions from other enterprises. Objective trust must be integrated with the current trust value, the past trust value, and the trust values provided by other enterprises. The integral trust evaluation method consists of the three steps discussed below. 6.1. Calculation of past trust value (PTV) (Step 1) The trust value decreases over time. Hence, this study adjusts the past trust value by applying a rate of decrease (Schonseleben, 2000). Two enterprises may have more than one past trust value. To evaluate the effect of past trust, a total past trust value is calculated using (E.15).

PM PTV ðTÞ ¼

ðt=hÞ PTV m¼1 ½e

M

;

ðE:15Þ

where PTV(T) is the total past trust value; PTV is a past trust value; e(t/h) is the rate of decrease of the past trust value; t is the interval of time since the past trust value was applied; h is the average rate of decrease of the past trust value; and M is the number of past trust values. 6.2. Calculation of others’ trust values (OTV) (Step 2)

wn ¼

N X 

 wneðt=hÞ OTV ;

n¼1 PTV nðTÞ PN n n¼1 PTV ðTÞ

;

6.3. Calculation of integral trust value (Step 3) The integral trust value (E.18) is calculated from current trust, past trust value, and others’ trust values from other enterprises and is obtained using Functions 14, 15, and 16, respectively. Function 18 must be constrained by (E.19), in which three weighting factors are determined by the VE leader, based on the importance of current trust, past trust, and the trust values provided by other enterprises.

TV ¼ wcCTRV þ wpPTVðadjÞ þ woOTVðadjÞ;

ðE:18Þ

wc þ wp þ wo ¼ 1;

ðE:19Þ

where wc, wp, and wo are the weights of current trust, past trust, and the others’ trust values provided by other enterprises, respectively. The trust level in Table 2 provides a reference when making knowledge-sharing decisions only by the enterprise that has the sought knowledge. 7. Simulation and verification of method This section verifies the proposed trust evaluation method for knowledge sharing, using a simplified, imagined VE case study as an example. This section includes an implementation of the proposed method using the tool Matlab. Fig. 5 displays the graphical user interface (GUI) simulated in Matlab to evaluate trust for VE knowledge sharing. Fig. 6 presents the process associated with activities of a VE. A worker of enterprise A is assigned to VER1 to execute activity A005, and a worker of enterprise C is assigned to VER5 to execute activity A007. Private knowledge associated with activity A005 is stored in enterprise A. When VER5 applies for enterprise A to share private knowledge concerning activity A005, enterprise A should evaluate its trust in enterprise C to determine whether that private knowledge should be shared. The method for evaluating trust between enterprises A and C is defined below. Table 3 presents information about knowledge related to activities A005 and A007. 7.1. Example calculation of activity correlation

Trust values provided by other enterprises may also decrease as time passes. The importance of the trust value provided by other enterprises varies with the degree of trust between the enterprise that provided the trust values and the enterprise of the owner of the desired knowledge. Therefore, a total OTV is calculated by (E.16) and (E.17).

OTV ðTÞ ¼

where OTV(T) is the total value of others’ trust, provided by other enterprises; OTV is the others’ trust value provided by other enterprises; e(t/h) is the rate of decrease of trust value; t is the interval of time since the past trust value was applied; h is the average rate of decrease of the trust value; N is the number of enterprises that provide opinions; wn is the weighting of the opinions of the enterprises; and PTV nðTÞ is the past trust value between the enterprise that provided trust values and the enterprise of the owner of the desired knowledge.

ðE:16Þ ðE:17Þ

(Step 1) Activity relation calculation Two activities, A005 and A007, are connected by two paths. The shortest distance between the activities is calculated using Dijkstra’s Algorithm (F.1):

Table 2 TV range and corresponding levels of knowledge sharing. TV

0 < TV < 0.25

0.25 < TV < 0.5

0.5 < TV < 0.75

0.75 < TV < 1

Trust Level

Very low

Low

Medium

High

861

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

Fig. 5. User interface used to evaluate trust for knowledge sharing.

VER5 A003 A001

0.4

0.4

0.5 0.4

Enterprise C A009 0.4

0.4

0.4 A002

0.4

A007 A005

0.4

0.5

A006

0.4 0.4

A004

A008

A010

0.3 0.4

VER1 Enterprise A Fig. 6. Example of the process associated with VE activities.

Table 3 Information about knowledge associated with activity. Activity ID

Instance

Property

A005

Sop052, C032, Sop012, T072, T055, C052, P455, P458 C021, Sop214, C032, Sop012, T055, P455, P456, P455

Characteristic, development, usage, maintain, physical, materials, machine, quality engineering

A007

Characteristic, development, cost accounting, statistics analysis, process analysis, quality engineering

dðA005; A007Þ ¼ min fdðA005; A006Þ þ wðA006; A007Þg u2S

v 2S

wðA006; A007Þ ¼ min fdðA006; A007Þ; dðA006; A008Þ þ wðA008; A007Þg u2S

v 2S

¼ minf0:4; 1:1g ¼ 0:4

The shortest distance between activity A005 and A007 is 0.8. Assume that the decision weight a set by the VE leader is 0.3. The strength of the relation between A005 and A007 is then calculated as follows (F.2):

ar A005A007 ¼

1 1 1 ¼ ¼ 0:8065: ¼ 1 þ adA005A007 1 þ ð0:3Þð0:8Þ 1:24

u2S

v 2S

dðA005; A007Þ ¼ minf0:4 þ 0:4g ¼ 0:8: u2S

v 2S

Consequently, the strength of the relation between A005 and A007 is 0.8065. (Step 2) Calculation of similarity between instances of activity knowledge

862

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

The similarity between instances of knowledge associated with activity A005 and those of A007 is calculated using the instance similarity calculation function (F.3) that was proposed in Section 4.2:

IA324 [ IA332 ¼ fC021; C032; C052; P455; P456; P458; Sop012;

Table 4 Information concerning past trust. h

1460

Time interval (Day) e(t/h) PTV

100 0.930 0.85

250 0.842 0.66

350 0.786 0.72

500 0.710 0.93

Sop052; Sop214; T055; T072g IA324 \ IA332 ¼ fC032; P455; Sop012; T055g InstanceSmðIA324 ; IA332 Þ ¼ 4=11 ¼ 0:3636: Therefore, the similarity between instances of knowledge associated with activity A005 and those of A007 is 0.3636. (Step 3) Calculation of similarity between properties of activity knowledge The property similarity calculation function (F.4), proposed in Section 4.3 is used to calculate the similarity between the properties of knowledge associated with activity A005 and those of activity A007, as follows: PA324 [ PA332 ¼ fmachine; characteristic; cost accounting;development; maintain materials;physical; process analysis; quality engineering; statistics analysis; usageg PA324 \ PA332 ¼ fcharacteristic; development; quality engineeringg PropertySmðPA324 ; PA332 Þ ¼ 3=11 ¼ 0:2727:

Therefore, the similarity between the properties of knowledge associated with the two activities is 0.2727. (Step 4) Calculation of the correlation between activities The activity correlation calculation function (F.5), proposed in Section 4.4, is adopted to calculate the correlation between activity A005 and A007. The three weighting factors in the activity relation (wr, wi, and wp) are assumed to be 0.4, 0.3, and 0.3, respectively. Therefore, the result is 0.6112.

ACA324

A332

¼ 0:6ð0:8065Þ þ 0:2ð0:3636Þ þ 0:2ð0:2727Þ ¼ 0:6112:

7.2. Example current trust evaluation (Step 1) Preprocessing input variables AC, CR, and CT are the input variables in the current trust evaluation method. The input value of AC is 0.6112, obtained from the above step. Based on the same assumptions made in Section 5, three collaboration relations are defined. First, the VE leader determines the fuzzy collaboration relations using E. 7:

e ¼ 0:15 þ 0:5 þ 0:85 : CR o col cor Assume that the relationship between two enterprises that separately execute activities A005 and A007 is collaborative. Hence, the input value of CR is 0.5. Assume also that these two enterprises must collaborate to execute the activities over 12 months, so the input value of CT is 12.

Table 5 Others’ trust values, provided by other enterprises. Enterprise h Time interval (Day) e(t/h) PTVn OTV

E5

E8

E9

250 0.842 0.8 0.48

1460 108 0.928 0.72 0.36

1000 0.504 0.57 0.65

0:930  0:85 þ 0:842  0:66 þ 0:786  0:72 þ 0:710  0:93 4 ¼ 0:643:

PTVðTÞ ¼

(Step 2) Calculation of others’ trust value Table 5 presents others’ trust values for enterprise C, provided by other enterprises. First, the weight of the opinion provided by each enterprise is calculated using (E.17) in Section 6.2. The results are as follows.

0:57 ¼ 0:2727 0:8 þ 0:72 þ 0:57 0:72 ¼ 0:3445 w2 ¼ 0:8 þ 0:72 þ 0:57 0:8 ¼ 0:3828 w5 ¼ 0:8 þ 0:72 þ 0:57

w1 ¼

Next, OTV is calculated using (E.18) in Section 6.3. The result is as follows:

OTVðadjÞ ¼ 0:2727  0:842  0:48 þ 0:3445  0:928  0:36 þ 0:3827  0:504  0:65 ¼ 0:3507 (Step 3) Calculation of the integral trust value Assume that the three trust weightings (wc, wp, and wo) are 0.6, 0.2, and 0.2, respectively. The integral trust value, calculated using (E.18), results in the following:

TV ¼ 0:6  0:6374 þ 0:2  0:643 þ 0:2  0:3507 ¼ 0:5812:

Finally, the trust level is medium, as indicated by Table 2, and provides a reference for knowledge-sharing decisions allowing enterprise A to determine whether the private knowledge associated with activity A005 should be shared with enterprise C. 8. Conclusions

(Step 2) Fuzzy inference The fuzzy inference engine is constructed using Matlab to evaluate the current trust between these two enterprises, with AC = 0.6112, CR = 0.5, and CT = 12 as the inputs. Fuzzy inference yields CTRV = 0.6374. 7.3. Example evaluation of integral trust (Step 1) Calculation of past trust value Table 4 presents information concerning the past trust between enterprises A and C. The past trust value is calculated using (E.15) in Section 6.1. The result is as follows.

In VEs, the willingness to share knowledge depends on a number of factors, including the information security technology infrastructure, the access control mechanism, the processes of the VE, and trust among allied enterprises. This study designs a trustbased, knowledge-sharing model stemming from the characteristics of VEs and designs a structure to express knowledge associated with VE activities. Finally, a fuzzy trust evaluation method for sharing knowledge derived from the activities of a VE and the interactions among allied enterprises, including collaborative relations and the period of collaboration, is developed. This approach consists of the evaluation of the correlation among activities, the eval-

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

uation of current trust, and the evaluation of integral trust. The proposed method considers the characteristics of trust, including its fuzziness, dynamism, and transitivity, among others, to evaluate an objective trust value. The purpose of the proposed method is to yield a trust level that can be used as a reference for making decisions about knowledge sharing. This method improves willingness to share knowledge and thereby increases the efficiency and competitiveness of VEs. Virtual enterprises reach their goals by using both organizational and employee knowledge. Knowledge stored in an employee’s brain is tacit knowledge, another important issue in the field of knowledge sharing. In the future, trust between workers in VEs will be evaluated using social network analysis (SNA) techniques and fuzzy theory so as to provide valuable knowledge that will help workers in VEs complete their tasks efficiently. Consequently, knowledge sharing will not only promote the efficient application of knowledge but will also promote the creation of new knowledge. Acknowledgement The authors would like to thank the National Science Council of the Republic of China, Taiwan, for financially supporting this research under Contract Nos. NSC98-2221-E-343-009, NSC98-2221E-006-232, and NSC98-2221-E-006-190. References Agnarsson, G., & Greenlaw, R. (2007). Graph theory: Modeling, applications, and algorithms. New Jersey: Prentice Hall. Au, R., Looi, M., & Ashley, P. (2001). Automated cross-organizational trust establishment on extranets. In Proceeding of workshop on information technology for virtual enterprise, Queensland, Australia, January 29th–30th (pp. 3–11). Bakker, M., Leenders, R. T. A. J., Gabbay, S. M., Kratzer, J., & Engelen, J. M. L. V. (2006). Is trust really social capital? Knowledge sharing in product development projects. The Learning Organization, 13(6), 594–605. Beckett, R. C. (2004). Exploring virtual enterprises using activity theory. Australasian Journal of Information Systems, 12(1), 103–110. Beckman, T. (1997). A methodology for knowledge management. In International Association of Science and Technology for Development (IASTED) AI and soft computing conference, Banff, Canada. Bharadwaj, K. K., & Al-Shamri, M. Y. H. (2009). Fuzzy computational models for trust and reputation systems. Electronic Commerce Research & Applications, 8, 37–47. Boon, S. D., & Holmes, J. G. (1991). The dynamics of interpersonal trust: Resolving uncertainty in the face of risk. In R. A. Hinde & J. Groebel (Eds.), Cooperation and prosocial behavior (pp. 190–211). UK: Cambridge University Press. Butler, J. K. (1991). Toward understanding and measuring conditions of trust: Evolution of the conditions of trust inventory. Journal of Management, 17, 643–663. Castelfranchi, C., Falcone, R., & Pezzulo, G. (2003). Integrating trustfulness and decision using fuzzy cognitive maps. Trust Management, 2692, 195–210. Chang, E. J., Hussain, F. K., & Dillon, T. S. (2005). Fuzzy nature of trust and dynamic trust modeling in service oriented environments. In Proceedings of the 2005 workshop on secure web services, Fairfax, VA, USA, November 11th (pp. 75–83). Chen, T. Y. (2008). Knowledge sharing in virtual enterprises via an ontology-based access control approach. Computers in Industry, 59(5), 502–519. Cheng, W., Hailin, L., & Hongming, X. (2008). Does knowledge sharing mediate the relationship between trust and firm performance? In Processing of international symposiums on information, Moscow, Russia, May 23th–25th (pp. 449–453). Chowdhury, S. (2005). The role of affect- and cognitions-based trust in complex knowledge sharing. Journal of Managerial Issues, 17(3), 310–326. Davenport, T. H., & Prusak, L. (1998). Working knowledge: How organization manage what they know. New York: Harvard Business School. Dodgson, M. (1993). Learning, trust and technological collaboration. Human Relations, 46(1), 77–96. Ellonen, R., Blomqvist, K., & Puumalainen, K. (2008). The role of trust in organisational innovativeness. Journal of Innovation Management, 11(2), 160–181. Fla’via Maria, S., Marcos, R. S., Borges, B., & Erick, A. R. (2006). Collaboration and knowledge sharing in network organizations. Expert Systems with Applications, 31, 715–727. Flaminio, T. G., Pinna, M., & Tiezzi, E. B. P. (2008). A complete fuzzy logical system to deal with trust management systems. Fuzzy Sets & Systems, 159(10), 1191–1207.

863

Ford, D. (2003). Trust and knowledge management: The seeds of success. In C. W. Holsapple (Ed.), Handbook on knowledge management (Vol. 1, pp. 553–576). New York: International Handbooks on Information Systems. Gollmann, D. (1999). Computer security. New York: John Wiley & Sons. Gruber, H. G. (2000). Does organizational culture affect the sharing of knowledge: The case of a department in a high-technology company. Ottawa, Ontario: Carleton University. Guha, S., Rastogi, R., & Shim, K. (1998). Cure: An efficient clustering algorithm for large database. In Proceedings of ACM SIGMOD international conference on management of data, Seattle, Washington, United States, June 1th–4th (pp. 73–84). Holland, D., & Reeves, J. R. (1994). Activity theory and the view from somewhere: Team perspectives on the intellectual work of programming. Mind, Culture, and Activity, 1(1 & 2), 8–24. Kanawattanachai, P., & Yoo, Y. (2002). Dynamic nature of trust in virtual teams. Journal of Strategic Information Systems, 11, 187–213. Kaptelinin, V., & Nardi, B. A. (1997). Activity theory: Basic concepts and applications. In Conference on human factors in computing systems (pp. 158–159). Kaptelinin, V., & Nardi, B. A. (2006). Acting with technology: Activity theory and interaction design. Cambridge, Mass: MIT Press. Koehn, D. (2003). The nature of and conditions for online trust. Journal of Business Ethics, 43, 3–19. Lawson, B., Petersen, K. J., Cousins, P. D., & Handfield, R. B. (2009). Knowledge sharing in interorganizational product development teams: The effect of formal and informal socialization mechanisms. Journal of Product Innovation Management, 26(2), 156–172. Lee, J. N. (2001). The impact of knowledge sharing, organizational capability and partnership quality on IS outsourcing success. Information & Management, 38(5), 323–335. Lewicki, R. J., & Bunker, B. B. (1996). Developing and maintaining trust in working relationships. In R. M. Kramer & T. R. Tyler (Eds.), Trust in organizations: Frontiers of theory and research (pp. 114–139). Thousand Oaks, CA: Sage Publications. Lin, C., & Tseng, S. M. (2005). The implementation gaps for the knowledge management system. Industrial Management & Data Systems, 105(2), 208–222. Lin, W. B. (2008). The effect of knowledge sharing model. Expert System with Applications, 34, 1508–1521. Lin, Y. C., Wang, L. C., & Tserng, H. P. (2006). Enhancing Knowledge exchange through web map-based knowledge management system in construction: Lessons learned in Taiwan. Automation in Construction, 15, 693–705. Ling, T. N., San, L. Y., & Hock, N. T. (2009). Trust: Facilitator of knowledge-sharing culture. Communications of the IBIMA, 7, 137–142. Luo, J., Liu, X., & Fan, M. (2009). A trust model based on fuzzy recommendation for mobile ad-hoc networks. Computer Networks, 53(14), 2396–2407. Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organisational trust. Academy of Management Review, 20(3), 709–734. McEvily, B., Perrone, V., & Zaheer, A. (2003). Trust as an organization principle. Organization Science, 14(1), 91–103. Meyerson, S., Weick, K. E., & Kramer, R. M. (1996). Swift trust and temporary groups. In R. M. Kramer & T. R. Tyler (Eds.), Trust in organizations: Frontiers of theory and research (pp. 166–195). Thousand Oaks, CA: Sage Publications. Mishira, A. K. (1996). Organizational responses to crisis: The centrality of trust. In R. M. Kramer & T. M. Tyler (Eds.), Trust in organizations (pp. 261–287). Newbury Park, CA: Sage. Mun, J., Shin, M., Lee, K., & Jung, M. (2009). Manufacturing enterprise collaboration based on a goal-oriented fuzzy trust evaluation model in a virtual enterprise. Computers & Industrial Engineering, 56, 888–901. Nardi, B. A. (1996). Context and consciousness: Activity theory and human–computer interaction. Massachusetts Institute of Technology. Negnevitsky, M. (2005). Artificial intelligence – A guide to intelligent systems. Boston: Addison-Wesley. Newell, S., David, G., & Chand, D. (2007). Exploring trust among globally distributed work teams. In Proceedings of the 40th Hawaii international conference on system sciences, January (p. 246). Big Island, Hawaii: IEEE Computer Society Press. Probst, G. J. B., Raub, S., & Romhardt, K. (2000). Managing knowledge: Building block for success. England: John Wiley & Sons Ltd. Quigley, N. R., Tesluk, P. E., Locke, E. A., & Bartol, K. M. (2007). A multilevel investigation of the motivational mechanisms underlying knowledge sharing and performance. Organization Science, 18(1), 71–88. Quinn, J. B., Anderson, P., & Finkelstein, S. (1996). Managing professional intellect: Making the most of the best. Boston: Harvard Business Review. Renzl, B. (2008). Trust in management and knowledge sharing: The mediating effects of fear and knowledge documentation. Omega, 36(2), 206–220. Riegelsberger, R., Sasse, M., & McCarthy, J. (2003). The researcher’s dilemma: Evaluating trust in computer-mediated communication. International Journal of Human–Computer Studies, 58, 759–781. Schmidt, S., Steele, R., Dillon, T. S., & Chang, E. (2007). Fuzzy trust evaluation and credibility development in multi-agent systems. Applied Soft Computing, 7, 492–505. Schonseleben, P. (2000). With agility and adequate partnership strategies towards effective logistics networks. Computers in Industry, 42(1), 33–42. Strader, T. J., Lin, F. R., & Shaw, M. J. (1998). Information infrastructure for electronic virtual organization management. Decision Support System, 23(1), 75–94. Victor, P., Cornelis, C., Cock, M. D., & Silva, P. P. (2009). Gradual trust and distrust in recommender systems. Fuzzy Sets and Systems, 160(10), 1367–1382. Wei, Z., Lu, L., & Yanchun, Z. (2008). Using fuzzy cognitive time maps for modeling and evaluating trust dynamics in the virtual enterprises. Expert Systems with Applications, 35, 1583–1592.

864

T.-Y. Chen et al. / Computers & Industrial Engineering 59 (2010) 853–864

Willem, A., & Buelens, M. (2007). Knowledge sharing in public sector organizations: The effect of organizational characteristics on interdepartmental knowledge sharing. Journal of Public Administration Research and Theory, 14(4), 581–606. Yang, Z., & Lin, H. (2008). Assessment of knowledge shared risk in virtual enterprise based on modified analytic hierarchy process. In Proceedings of the 2008 IEEE,

international conference on information and automation, Zhangjiajie, China, June 20th–23th (pp. 1056–1060). Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8, 338–353. Zolin, R., Hinds, P. J., Fruchter, R., & Levitt, R. E. (2004). Interpersonal trust in cross functional, geographically distributed work: A longitudinal study. Information and Organization, 14, 1–26.