You are on page 1of 12

ARTICLE IN PRESS

Int. J. Production Economics 124 (2010) 159–170

Contents lists available at ScienceDirect

Int. J. Production Economics


journal homepage: www.elsevier.com/locate/ijpe

A method for partner selection of codevelopment alliances using individual


and collaborative utilities
Bo Feng a,b,c, Zhi-Ping Fan a,, Jian Ma c
a
Department of Management Science and Engineering, School of Business Administration, Northeastern University, Shenyang 110004, China
b
Department of Decision Science, School of Business Administration, South China University of Technology, Guangzhou 510640, China
c
Department of Information Systems, City University of Hong Kong, Kowloon, Hong Kong, China

a r t i c l e in fo abstract

Article history: The success of a codevelopment alliance depends on the close cooperation between partners.
Received 18 March 2008 Consideration of collaborative utility is very important for partner selection in the alliance formation,
Accepted 5 October 2009 while it is neglected in the existing research. The purpose of this paper is to propose a method for
Available online 28 October 2009
partner selection of codevelopment alliances using individual and collaborative utilities. Firstly, a
Keywords: framework for partner selection of codevelopment alliances is presented. It describes two types of
Partner selection attributes, individual attributes and collaborative ones, as well as their measurements. According to the
Codevelopment alliance framework, a fuzzy multiple attribute decision-making (FMADM) approach is then proposed to
Individual utility integrate the assessment data of individual and collaborative utilities to achieve the final ranking of all
Collaborative utility
candidate partners. Additionally, an example is used to illustrate the potential application of the
Fuzzy multiple attribute decision-making
proposed method.
(FMADM)
& 2009 Elsevier B.V. All rights reserved.

1. Introduction In the existing research on partner selection, the individual


utility of each candidate partner is focused on, but the
In an increasingly competitive global market, new product collaborative utility shared by pairwise partners, is overlooked.
development (NPD) becomes a key source for firms to advance For example, in supplier selection, firms focus on the individual
core competence advantage (Emden et al., 2006; Afonso et al., utility of each candidate supplier, such as quality, availability,
2008). A firm survives by having good contacts with other firms reliability and performance (Vonderembse and Tracey, 1999). In
who hold complementary assets (Cowan et al., 2007). Codevelop- fact, consideration of collaborative utility is very important in
ment alliance is usually adopted in the new-generation product partner selection since it may reduce the uncertainty of future
development (Emden et al., 2006). It is also a NPD strategy for cooperation between the selected partners. As for partner
firms to improve quality, cut cost, short product developing time selection of codevelopment alliances, the good collaboration
and share risk, as well as achieve a goal which cannot be realized situation between partners contains the noncompeting goal,
by each firm’s single effort. Thus, more and more firms have been compatible cultures and so on (Emden et al., 2006), all of which
involved in different codevelopment alliances at the same time. avail to the future communication, knowledge sharing and
Although more and more firms have devoted into codevelop- reciprocal exchange of information between the selected partners.
ment alliances in the past decade, a significant number of Meanwhile, good collaboration situation between desired part-
alliances failed (Dacin et al., 1997; Arino, 2003). Besides the ners allows for cross-disciplinary integration, which may be
inherent risk, one of the most important reasons is the essential for creating new products (Chesbrough, 2003; Feng
incompatibility between partners (Büyüközkan et al., 2008). et al., 2009; Fan, et al., 2009). It may lead to shared research and
Partner selection becomes a strategic problem for firms (Cowan development (R&D), reduce costs and risks (Perks, 2000), create
et al., 2007). It may affect an alliance performance and even opportunities for the utilization of technologies that have not yet
decide the alliance’s fate. It is therefore an important decision been found application, or increase the speed to a new market
problem in the formation of codevelopment alliances, but few (Bronder and Pritzl, 1992; Deck and Strom, 2002). Therefore, the
scholars pay much attention to it. collaborative utility should be considered in partner selection of
codevelopment alliances. Thus, effective methods should be
investigated to measure the collaborative utility between partners
and integrate individual and collaborative utilities to achieve the
 Corresponding author. Tel.: + 86 24 8368 7753; fax: + 86 24 2389 1569. final ranking of candidate partners.
E-mail addresses: neu_fengbo@163.com (B. Feng), zpfan@mail.neu.edu.cn This paper proposes a method for partner selection of codevelop-
(Z.-P. Fan). ment alliances using individual and collaborative utilities. Firstly, a

0925-5273/$ - see front matter & 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpe.2009.10.019
ARTICLE IN PRESS
160 B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170

framework for partner selection is presented, in which both strategic alliances in logistics chains. The criteria include similar
individual utility and collaborative utility are considered. The values-goals, similar size, financial stability, comparable cultures,
attributes to measure the two types of utilities for partner selection successful track record, sustainable relationship, technical ex-
of codevelopment alliances is finalized based on literature review, pertise, performance, market knowledge and managerial experi-
which can be classified into individual attributes and collaborative ence. Araz and Ozkarahan (2007) pointed out that traditional
ones. With respect to the two kinds of attributes, the individual selection criteria must not be solely used for strategic supplier
utility matrix and collaborative utility matrix are constructed, where selection and evaluation, such as cost, quality and delivery. In
elements of the two matrices are in the form of linguistic terms. strategic sourcing, many other criteria should be considered with
Then, using Fuzzy Set Theory, the linguistic terms are expressed as the aim of developing a long-term supplier relationship, such as
triangular fuzzy numbers, and a fuzzy multiple attribute decision- quality management practices, long-term management practices,
making (FMADM) approach is proposed to respectively compute the financial strength, technology and innovativeness level, suppliers’
overall assessment values of individual utility and collaborative cooperative attitude, supplier’s co-design capabilities, and cost
utility. Furthermore, the assessment values of individual utility and reduction capabilities.
collaborative utility are integrated to obtain the overall assessment The above literature significantly contributes to our research,
value of each candidate partner. Finally, the ranking order of all especially the work of Emden et al. (2006). In Emden et al.’s
candidate partners is obtained according to the derived overall theoretical model, collaborative factors are referred, which
assessment values. provides us with a new perspective that collaborative utility
The rest of the paper is organized as follows. Section 2 reviews should be considered in partner selection of codevelopment
the literature on partner selection of alliances. Section 3 presents a alliances. However, in the existing research, the individual utility
framework for partner selection. Section 4 proposes a FMADM of each candidate partner is mostly used, while the collaborative
approach using individual and collaborative utilities to select the utility shared by pairwise partners has seldom been a focus.
desirable partners for a codevelopment alliance. Section 5 describes On the other hand, there has been much research addressed
the potential application of the proposed method. Finally, Section 6 the methods for partner selection. The methods can be classified
contains some conclusions and the suggested future works. into three categories: mathematical programming approaches
(Zeng et al., 2006; Hajidimitriou and Georgiou, 2002; Amid et al.,
2006), rating/linear weighting approaches (Wang and Chen, 2007;
2. Literature review Ding and Liang, 2005; Chang et al., 2006; Saen, 2007; Huang et al.,
2004) and artificial intelligence techniques (Fischer et al., 2004).
So far, research on partner selection of codevelopment These approaches or techniques can be used exclusively or in a
alliances is limited, thus the broad and indirectly literature on hybrid way. Table 1 shows the partner selection approaches or
partner selection of alliances are reviewed. Partner selection of techniques in methodological area. The individual utility
alliances is a crucial strategic decision problem for firms. The considered in the existing methods is associates with a single
literature on it is sharply increasing in the last decade (Verma and candidate partner. However, the collaborative utility additionally
Pullman, 1998; Geringer, 1988; Vonderembse and Tracey, 1999; considered in this paper is shared by pairwise partners.
Mikhailov, 2002; Hajidimitriou and Georgiou, 2002; Huang et al., Furthermore, the aggregation of collaborative utility between
2004; Fischer et al., 2004; Ding and Liang, 2005; Amid et al., 2006; pairwise partners differs from that of a single candidate partner.
Chang et al., 2006; Wang and Chen, 2007; Zeng et al., 2006; Saen, Additionally, individual and collaborative utilities should be
2007; Büyüközkan et al., 2008). The existing studies mainly focus integrated to obtain the overall ranking value of each candidate
on partner selection for virtual enterprises (Wang and Chen, partner. Therefore, existing methods could not be directly used to
2007; Zeng et al., 2006), partner selection for dynamic alliances solve the problem addressed in this paper. A novel method needs
(Ding and Liang, 2005), partner selection for international joint to be investigated for partner selection of codevelopment
ventures (Chang et al., 2006; Hajidimitriou and Georgiou, 2002) alliances using individual and collaborative utilities.
partner selection for production networks (Huang et al., 2004;
Fischer et al., 2004), and supplier selection in supply chains (Amid
et al., 2006; Saen, 2007; Famuyiwa et al., 2008). 3. A framework for partner selection of codevelopment
The attributes (or criteria) for partner selection of alliances alliances
have been discussed in many studies. Herein, Verma and Pullman
(1998) ranked the importance of the attributes for supplier In this section, an evaluation hierarchy for partner selection of
selection, i.e., quality, on-time delivery, cost, lead-time and codevelopment alliances is constructed, as shown in Fig. 1. The
flexibility. Geringer (1988) found that a partner’s culture, past hierarchy involves two types of attributes, individual attributes
experience, size and structure are as important in partner and collaborative attributes, all of which are finalized according to
selection as task-related criteria, such as financial assets, manage- the aforementioned literature and the real requirements of the
rial experience, access to markets and so on. Vonderembse and formation of codevelopment alliances. The individual attributes
Tracey (1999) regarded that supplier could be selected by quality, include technology capability, financial health, knowledge and
availability, reliability and performance. Mikhailov (2002) em- managerial experience, and capability to access new market. The
ployed price of the product, quality of the product, financial collaborative attributes consist of resource complementarity,
stability and quality of customer service to select partnership in overlapping knowledge bases, motivation correspondence, goal
the formation of virtual enterprises. Emden et al. (2006) corresponded and compatible cultures. Brief descriptions for
developed a process theoretical model for partner selection of these attributes are presented in Tables 2 and 3.
collaborative NPD alliances. In their theoretical model, technical In Tables 2 and 3, all attributes for partner selection are
ability, resource complementarity, overlapping knowledge bases, qualitative, so it is convenient for experts to express their
motivation correspondence, goal correspondence, compatible opinions using linguistic terms (namely, linguistic labels) such
cultures, propensity to adapt ling-tem orientation are considered as ‘‘Very high’’, ‘‘High’’ or ‘‘Middle’’. Therefore, the situation that
for partner selection in different alliance phases. Büyüközkan experts express their opinions using linguistic terms is considered
et al. (2008) used two dimensions, strategy and business in this study. Furthermore, the measurements of the two types of
excellence, and 10 extended criteria for partner selection of attributes are different due to their characteristics. Measurement
ARTICLE IN PRESS
B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170 161

Table 1
Methods for partner selection of alliances.

Relevant studies Methods Data formats

Zheng et al. (2006) Nonlinear integer programming Crisp numbers


Applied Mathematics and Computation
Hajidimitriou and Georgiou (2002) Goal programming Crisp numbers
European Journal of Operational Research
Amid et al. (2006) Fuzzy multi-objective linear Crisp numbers
International Journal of Production Economics programming
Wang and Chen (2007) Fuzzy preference programming Interval numbers
Omega Consistent fuzzy preference relations
Ding and Liang (2005) Fuzzy MCDM Fuzzy linguistic terms
Information Sciences
Chang et al. (2006) Fuzzy MADA Fuzzy linguistic terms
International Journal of Production Economics Crisp numbers
Saen (2007) DEA Cardinal and ordinal data
European Journal of Operational Research
Huang et al. (2004) MCDM Crisp numbers
International Journal of Computer Integrated
Manufacturing
Fischer et al. (2004) Intelligent optimization Crisp number
Robotics and Computer-Integrated Manufacturing

Technology capability (I1)

Financial health (I2)


Individual utility
Knowledge and managerial experience (I3)

Capability to access new market (I4)

Goal
Resource complementarity (C1)

Overlapping knowledge bases (C2)

Collaborative utility Motivation correspondence (C3)

Goal correspondence (C4)

Compatible cultures (C5)

Fig. 1. The evaluation hierarchy for partner selection of codevelopment alliances.

Table 2
Definitions of individual attributes.

Individual attributes Definitions

Technology capability (I1) The partner has either an innovative technology or expertise in a certain field
Financial health (I2) The partner has good financial status on return on stockholders’ equity, return on assets, return on long-term investment, and profit
margin
Knowledge and managerial The partner has either resourceful knowledge or managerial experience of NPD
experience (I3)
Capability to access new The partner has good relationship with local government, understands competitors and customers, and can provide high quality
markets (I4) service

Table 3
Definitions of collaborative attributes.

Collaborative attributes Definitions

Resource complementarity (C1) The partners have technical resource that is distinct yet complementing one another for the opportunity foreseen
Overlapping knowledge bases (C2) The partners have the similar knowledge bases, which allow them to predict the potential business opportunities
Motivation correspondence (C3) This refers to the extent to which the partners’ perceived ulterior motives are in correspondence with one another
Goal correspondence (C4) The partners have noncompeting goals, high goal correspondence that enhances the consistency of expectations and
assures mutual gains
Compatible cultures (C5) The partners have the compatible cognitions, expectations, mindsets, norms and values among different organizations
ARTICLE IN PRESS
162 B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170

Partner selection
Objective

Individual utility Collaborative utility Sub-objectives

I1 I2 I3 I4 C1 C2 C3 C4 C5 Attributes

P1 P2 P1 P2
P1 P3 P5 Candidate
P6 P* P3 P6 P3
P2 P4 P6 partners
P5 P4 P5 P4

Fig. 2. A framework for partner selection of codevelopment alliances.

of individual attribute is by means of experts’ assessment to each  ˆ (x)


candidate partner. Measurement of collaborative attributes is A

sorted on experts’ judgment to pairwise partners. Due to the 1


different measurements, aggregation approaches for two types of
attribute data are distinct.
A framework is presented for solving the problem of partner
selection for codevelopment alliances using individual and
collaborative utilities, as shown in Fig. 2. In the framework, an
initiator ðP Þ and six candidate partners ðP1 ; P2 ; . . . ; P6 Þ are taken as 0 L M R x
d d d
an example to illustrate the current problem.
The framework in Fig. 2 consists of four layers, i.e., objective, ^
Fig. 3. A triangular fuzzy number A.
sub-objectives, attributes and candidate partners. Brief descrip-
tions for each layer are given below.
fuzzy ratings (Li and Yang, 2004; Costantino and Gravio, 2009).
 Objective: it is to select the desired partners for a codevelop- Fuzzy Set Theory, initially introduced by Zadeh (1965), is very
ment alliance from a candidate partner set. helpful to deal with the vagueness of language in decision-
 Sub-objectives: two sub-objectives are considered to select the making. As for decision-making problems in a fuzzy environment
desired partners, one is individual utility and the other is to deal with linguistic terms provided by experts, it is better to
collaborative utility. convert the linguistic terms into fuzzy numbers (Chen, 1996; Lin
 Attributes: four individual attributes (see Table 2) are and Wu, 2008). In the following, we briefly introduce some
employed to measure individual utility, while five collabora- essential definitions of Fuzzy Set Theory (Zadeh, 1965; Kaufmann
tive ones (see Table 3) are adopted to measure collaborative and Gupta, 1985; Chen, 1996; Lin and Wu, 2008).
utility. The definitions of the two types of attributes are shown
in Tables 2 and 3. Definition 4.1. A fuzzy set A^ in a universe of discourse X is
 Candidate partners: it points the six candidate partners characterized by a membership function mA^ ðxÞ which associates
ðP1 ; P2 ; . . . ; P6 Þ. It needs to be emphasized that P represents with each element x in X, a real number in the interval [0,1]. The
the initiator for the formation of an alliance and it does not ^
function value mA^ ðxÞ is termed the grade of membership of x in A.
include in the candidate partner set. In addition, this layer also
shows the information measurement of individual and Definition 4.2. A fuzzy set A^ of the universe of discourse X is
collaborative attributes. In the left rectangle, the measurement called a normal fuzzy set implying that (xi A X; mA^ ðxi Þ ¼ 1.
of individual attributes concerning each candidate partner is
shown. In the right rectangle, the measurement of collabora- Definition 4.3. A fuzzy number A^ is called positive fuzzy number
tive attributes concerning pairwise partners is displayed. The if mA^ ðxi Þ ¼ 0 for all xi o 0.
star figure shows the measurement of the collaborative utility
between each candidate partner and initiator P . Meanwhile, Definition 4.4. A triangular fuzzy number A^ can be defined as a
the network figure shows the measurement of the collabora- triplet ðdL ; dM ; dR Þ (see Fig. 3). The membership function mA^ ðxÞ is
tive utility between pairwise candidate partners. defined as
8
>
> 0; x o dL ;
>
>
4. Fuzzy Set Theory < ðx  dL Þ=ðdM  dL Þ; dL r x rdM ;
mA^ ðxÞ ¼
> ðdR  xÞ=ðdR  dM Þ; dM rx r dR ;
>
>
>
Fuzziness is inherent in decision information and group : 0; x 4 dR ;
decision-making processes, and linguistic variables are well
suited to assessing an alternative on qualitative attributes using where dL , dM and dR are real numbers, such that dL rdM rdR .
ARTICLE IN PRESS
B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170 163

Theorem 4.1. Let A^ 1 ¼ ðdL1 ; dM R ^ L M R


1 ; d1 Þ and A 2 ¼ ðd2 ; d2 ; d2 Þ be two and collaborative attribute set, where Ii represents the ith
arbitrary triangular fuzzy numbers. The addition operation of A^ 1 and individual attribute and Cj represents the jth collaborative one.
Let S ¼ fsr jr ¼ 1; . . . ; gg be a pre-established finite and totally
A^ 2 , denoted by A^ 1  A^ 2 , yields another triangular fuzzy number, i.e.,
ordered linguistic term set with odd cardinalities, where sr is the
A^ 1  A^ 2 ¼ ðdL1 þdL2 ; dM M R R
1 þ d2 ; d1 þ d2 Þ. Here, notation  denotes fuzzy rth linguistic term and presents the words in natural language.
number additive operation. Here, we employ a linguistic term set with seven-ranking scale
according to real requirements, i.e., S= {s0 = DL: definitely low,
Theorem 4.2. Let A^ ¼ ðdL ; dM ; dR Þ be a triangular fuzzy number and
s1 = VL: very low, s2 = L: low, s3 = M: medium, s4 =H: high, s5 = VH:
k 40 a crisp number, then k  A^ ¼ ðkdL ; kdM ; kdR Þ. Here, notation 
very high, s6 =DH: definitely high}.
denotes fuzzy number multiplicative operation.
Suppose W ~ k ¼ ðw ~ 1k ; w ~ mk Þ and V~ k ¼ ðv~ 1k ; v~ 2k ; . . . ; v~ nk Þ
~ 2k ; . . . ; w
Theorem 4.3. Let A^ 1 ¼ ðdL1 ; dM R ^ L M R
1 ; d1 Þ and A 2 ¼ ðd2 ; d2 ; d2 Þ be two
respectively be individual attribute weight vector and collabora-
triangular fuzzy numbers, where dL1 ; dL2 Z0. The multiplication opera- tive attribute weight vector provided by expert Ek , where w ~ ik and
tion of A^ 1 and A^ 2 , denoted by A^ 1  A^ 2 , approximates a triangular fuzzy v~ jk are linguistic assessments of the importance of individual
^ ^ attribute Ii and collaborative attribute Cj , respectively, w ~ ik ; v~ jk A S.
number ðdL1 dL2 ; dM M R R L L M M R R
1 d2 ; d1 d2 Þ, i.e., A 1  A 2 ffi ðd1 d2 ; d1 d2 ; d1 d2 Þ (van
In this paper, we assume that all experts’ assessments on
Laarhoven and Pedrycz, 1983).
attribute weights, individual and collaborative attribute ratings
Definition 4.5. Linguistic variables are used as variables whose are in the form of linguistic terms. Note d~ as the linguistic
hik
values are not numbers but linguistic terms (Zadeh, 1975). assessment of candidate Ph with regard to individual attribute Ii
The linguistic variables are usually employed by decision- provided by expert Ek , d~ hik A S. Then the individual attribute
makers to express their opinions, which is very useful in dealing ratings provided by expert Ek can be represented by an individual
with ill-defined situations in traditional qualitative expressions. ~ k as follows:
utility matrix D
Linguistic terms can be represented into fuzzy numbers. Com-
I1 I2 Im
monly, the triangular fuzzy numbers are used. 2 3
Moreover, to obtain the ranking order of candidate partners, P1 d~ 11k d~ 12k d~ 1mk
6 7
usually, we had to convert the final fuzzy data into a crisp value. P2 6 ~
6 d 21k d~22k d~ 2mk 7
7
~ k ¼ ½d~ hik 
D ¼ 6 7; k ¼ 1; . . . ; l:
Opricovic and Tzeng (2003) proposed the converting fuzzy data
q
m ^ 6 ^ ^ ^ 7
4 5
into crisp scores (CFCS) method for defuzzification. This method Pq d~ q1k d~ q2k d~ qmk
has the advantages that a greater crisp value is associated with a
Furthermore, the linguistic assessments between the initiator
greater membership function and two symmetrical triangular
and candidate partners concerning collaborative attributes pro-
fuzzy numbers with the same mean can be distinguished.
vide by each expert is expressed by a vector X~ ¼ ðx~ ; x~ ;...; jk 11jk 22jk
Theorem 4.4. (The CFCS method). Let A^ k ¼ ðdLk ; dM R
k ; dk Þ be a positive x~ qqjk ÞT , where x~ hhjk denotes the judgment on collaborative utility
triangular fuzzy number, k ¼ 1; . . . ; n, and its representing crisp value between initiator P  and candidate Ph with regard to collaborative
Adef
k
can be obtained by attribute Cj provided by expert Ek , x~ hhjk A S, j ¼ 1; . . . ; n, k ¼ 1; . . . ; l,
M 2 2 L 2
h ¼ 1; . . . ; q. The collaborative attribute ratings provide by each
D½ðdM R L R M
k  LÞðD þ dk  dk Þ ðR  dk Þ þ ðdk  LÞ ðD þ dk  dk Þ 
Adef
k
¼ Lþ ; expert is represented by matrix Y~ jk , i.e.,
ðD þ dM
k
 dLk ÞðD þ dRk  dM
k
Þ2 ðR  dLk Þþ ðdRk  LÞðD þ dM
k
 dLk Þ2 ðD þ dRk  dM
k
Þ
ð1Þ P1 P2 Pq
2 3
where L ¼ minfdL1 ; dL2 ; . . . ; dLn g, R ¼ maxfdR1 ; dR2 ; . . . ; dRn g and D ¼ R  L. P1  y~ 12ik y~ 1qik
P2 6
6 y~ 21ik  y~ 2qik 7
7
Y~ jk ¼ ½y~ hfjk q
q ¼ 6 7;
5. The FMADM approach ^ 64 ^ ^ ^ 7 5
Pq y~ q1ik y~ q2ik 
Based on the framework for partner selection of codevelop- j ¼ 1; . . . ; n; k ¼ 1; . . . ; l; h; f ¼ 1; . . . ; q; ha f ;
ment alliances presented in Section 3, we propose a FMADM
where y~ hfjk denotes the judgment on the collaborative utility
approach using individual and collaborative utilities in this
section. In the proposed approach, firstly, linguistic terms are between candidates Ph and Pf concerning attribute Cj provided by
expressed as triangular fuzzy numbers. Then, the fuzzy numbers expert Ek , y~ hfjk A S.
are aggregated to obtain the overall assessment value of each For the convenience of analysis, let x~ hhjk ¼ c~ hhjk and y~ hfjk ¼ c~ hfjk ,
candidate partner. Furthermore, the overall assessment values in then vector X~ jk can be incorporated into matrix Y~ jk to constructed
the form of triangular fuzzy numbers are mapped into the crisp the collaborative utility matrix C~ jk :
numbers to achieve the final ranking of all candidate partners. The
P1 P2 Pq
details of the proposed approach are presented as follows. 2 3
P1 c~ 11jk c~ 12jk ~c 1qjk
P2 6
6 c~ 21jk c~ 22jk c~ 2qjk 7
7
5.1. Problem description C~ jk ¼ ½c~ hfjk q
q ¼ 6 7;
^ 6
4 ^ ^ ^ ^ 7 5
Let P ¼ fPh jh ¼ 1; . . . ; q; q Z2g be a finite candidate partner set, Pq c~ q1jk c~ q2jk c~ qqjk
where Ph is the hth candidate. It needs to be emphasized that j ¼ 1; . . . ; n; k ¼ 1; . . . ; l:
initiator P* is not included in set P since the initiator is selected
In matrix C~ jk , the assessment value of collaborative utility
surely. Let E ¼ fEk jk ¼ 1; . . . ; l; l Z2g be a finite expert set, where Ek
between candidates Ph and Pf is equal to the one between
is the kth expert who is invited by the decision-maker to conduct
the partner selection. Here, we suppose that the importance candidates Pf and Ph due to the characteristics of collaborative
degrees or weights of experts are the same. Let I ¼ fIi ji ¼ 1; . . . ; mg attributes in Table 3, i.e., c~ ¼ c~ . Thus, matrix C~
hfjk is a
fhjk jk
and C ¼ fCj jj ¼ 1; . . . ; ng respectively be the individual attribute set symmetrical matrix.
ARTICLE IN PRESS
164 B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170

In summary, the problem addressed in this paper is to select ^ with vector W


Aggregate matrix D ^ to obtain the overall
the most desirable partner(s) from set P using individual utility ^ Ind ¼ ðd^ 1 ; d^ 2 ; . . . ; d^ q Þ,
assessment vector of individual utility D
matrix D ~ k ¼ ðw
~ k , individual attribute weight vector W ~ 1k ; w
~ 2k ; . . . ;
where d^ denotes the overall assessment value of individual
h
~ mk Þ, collaborative utility matrix C~ jk and collaborative attribute
w utility of candidate Ph , and it is expressed as
weight vector V~ k ¼ ðv~ 1k ; v~ 2k ; . . . ; v~ nk Þ.
d^ h ¼ ð1=mÞ

5.2. Assessment of individual utility ^ 1 Þ  ðd^ h2  w


 ½ðd^ h1  w ^ 2 Þ   ðd^ hm  w
^ m Þ; h ¼ 1; . . . ; q: ð7Þ

The individual utility of each candidate partner is assessed in If we note that d^ h ¼ ðdLh ; dM R L M R
h ; dh Þ, then dh , dh and dh are respectively

this section. To process the linguistic terms, we consider calculated by


Pm
transforming the linguistic terms into triangular fuzzy numbers. wL dL
The linguistic term sr ðsr A SÞ can be represented by a triangular dLh ¼ i ¼ 1 i hi ; h ¼ 1; . . . ; q; ð8aÞ
m
fuzzy number by the following formula:
     Pm
r1 r r þ1 i¼1 wM M
i dhi
d^ ¼ ðdL ; dM ; dR Þ ¼ max ; 0 ; ; min ;1 ; ð2Þ dM
h ¼ ; h ¼ 1; . . . ; q; ð8bÞ
6 6 6 m
Pm
where r ¼ 0; 1; . . . ; 6. i¼1 wRi dRhi
dRh ¼ ; h ¼ 1; . . . ; q: ð8cÞ
Now, corresponding relations between linguistic variables and m
triangular fuzzy numbers can be obtained using Eq. (2), as shown
in Table 4. From this, W ~ k ¼ ðw
~ 1k ; w
~ 2k ; . . . ; w
~ mk Þ can be represented
5.3. Assessment of collaborative utility
in the form of triangular fuzzy number as W ^ k ¼ ðw
^ 1k ; w
^ 2k ; . . . ;
w ^ ik ¼ ðwLik ; wM
^ mk Þ, where w R ~ ~ Using the collaborative utility matrix C~ jk and collaborative
ik ; wik Þ. Likewise, D k ¼ ½d hik q
m can be
represented in the form of triangular fuzzy numbers as D ^k¼ attribute weight vector V~ k ¼ ðv~ 1k ; v~ 2k ; . . . ; v~ nk Þ, the overall assess-
ment value of collaborative utility of each candidate partner can
½d^ hik q
m , where d^ hik ¼ ðdLhik ; dM R
hik ; dhik Þ. be obtained.
Aggregate the individual attribute weight vector provided by
^ ¼ ðw^ 1; V~ k ¼ ðv~ 1k ; v~ 2k ; . . . ; v~ nk Þ is represented in the form of triangular
each expert into the collective attribute weight vector W
w ^ m Þ, where w
^ 2; . . . ; w ^ i is obtained by fuzzy numbers as V^ k ¼ ðv^ 1k ; v^ 2k ; . . . ; v^ nk Þ using Eq. (2), where
v^ jk ¼ ðvLjk ; vM R
jk ; vjk Þ. Likewise, the collaborative utility matrix
w ^ i1  w
^ i ¼ ð1=lÞ  ½w ^ i2   w
^ il ; i ¼ 1; . . . ; m: ð3Þ
~
C ¼ ½c~  is also represented in the form of triangular fuzzy
jk hfjk q
q
If we note that w^ i ¼ ðwLi ; wM R
i ; wi Þ, then there are
Pl numbers as C^ jk ¼ ½c^ hfjk q
q using Eq. (2), where c^ hfjk ¼ ðchfjk
L
;
wL M R
wLi ¼ k ¼ 1 ik ; i ¼ 1; . . . ; m; ð4aÞ chfjk ; chfjk Þ.
l
The collective weight vector V^ ¼ ðv^ 1 ; v^ 2 ; . . . ; v^ n Þ is obtained by
Pl M
k ¼ 1 wik aggregating collaborative attribute weight vectors V^ 1 ; V^ 2 ; . . . ; V^ l ,
wM
i ¼ ; i ¼ 1; . . . ; m; ð4bÞ
l where v^ j is represented by
Pl v^ j ¼ ð1=lÞ  ½v^ j1  v^ j2   v^ jl ; j ¼ 1; . . . ; n: ð9Þ
wRik
k¼1
wRi ¼ ; i ¼ 1; . . . ; m: ð4cÞ
l If we note that v^ j ¼ ðvLj ; vM R
j ; vj Þ, then there are
Aggregate the linguistic assessment provided by each expert
^ ¼ ½d^ hi  Pl
into the collective assessment and matrix D q
m is conse- k¼1 vLjk
quently obtained, where d^ hi is calculated by vLj ¼ ; j ¼ 1; . . . ; n; ð10aÞ
l
d^ hi ¼ ð1=lÞ  ½d^ hi1  d^ hi2   d^ hil ; h ¼ 1; . . . ; q; i ¼ 1; . . . ; m: ð5Þ Pl
k¼1 vM
jk
If we note that d^ hi ¼ ðdLhi ; dM R vM
j ¼ ; j ¼ 1; . . . ; n; ð10bÞ
hi ; dhi Þ,
then there are l
Pl
dL Pl
dLhi ¼ k ¼ 1 hik ; h ¼ 1; . . . ; q; i ¼ 1; . . . ; m; ð6aÞ k¼1 vRjk
l vRj ¼ ; j ¼ 1; . . . ; n: ð10cÞ
l
Pl
k¼1 dM Aggregate the collaborative utility matrices C^ j1 ; C^ j2 ; . . . ; C^ jl into
dM
hi ¼
hik
; h ¼ 1; . . . ; q; i ¼ 1; . . . ; m; ð6bÞ
l
the collective collaborative utility matrix C^ ¼ ½c^  , j ¼ 1; . . . ; n, j hfj q
q
Pl R where c^ hfj is obtained by
k ¼ 1 dhik
dRhi ¼ ; h ¼ 1; . . . ; q; i ¼ 1; . . . ; m: ð6cÞ
l c^ hfj ¼ ð1=lÞ
 ½c^ hfj1  c^ hfj2   c^ hfjl ; j ¼ 1; . . . ; n; h; f ¼ 1; . . . ; q: ð11Þ
Table 4
L M R
Linguistic terms and corresponding triangular fuzzy numbers. If we note that c^ hfj ¼ ðchfj ; chfj ; chfj Þ, then there are
Pl L
Linguistic terms Triangular fuzzy numbers L k ¼ 1 chfjk
chfj ¼ ; j ¼ 1; . . . ; n; h; f ¼ 1; . . . ; q; ð12aÞ
l
S0 = definitely low (DL) (0, 0, 0.17)
S1 = very low (VL) (0, 0.17, 0.33) Pl M
S2 = low (L) (0.17, 0.33, 0.5) M k ¼ 1 chfjk
chfj ¼ ; j ¼ 1; . . . ; n; h; f ¼ 1; . . . ; q; ð12bÞ
S3 = medium (M) (0.33, 0.5, 0.67) l
S4 = high (H) (0.5, 0.67, 0.83)
S5 = very high (VH) (0.67, 0.83, 1)
Pl R
R k ¼ 1 chfjk
S6 = definitely high (DH) (0.83, 1, 1) chfj ¼ ; j ¼ 1; . . . ; n; h; f ¼ 1; . . . ; q: ð12cÞ
l
ARTICLE IN PRESS
B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170 165

Aggregate the collective collaborative utility matrices C^ 1 ; C^ 2 ; . . . ; all candidate partners can be obtained and the desired partners
for a codevelopment alliance can be selected.
C^ n with the collaborative attribute weight vector V^ ¼ ðv^ 1 ; v^ 2 ; . . . ; v^ n Þ
In summary then, the procedure of the proposed FMADM
to obtain the overall collaborative utility matrix C^ ¼ ½c^ hf q
q , where
approach is:
c^ hf denotes the overall assessment value of collaborative utility Step 1: Obtain the individual attribute weight vector
between candidates Ph and Pf , and it can be expressed as W ~ k ¼ ðw
~ 1k ; w
~ 2k ; . . . ; w
~ mk Þ, the collaborative attribute weight vector
c^ hf ¼ ð1=nÞ  ½ðc^ hf 1  v^ 1 Þ V~ ¼ ðv~ ; v~ ; . . . ; v~ Þ, the individual utility matrix D ~ ¼ ½d~ k
1k 2k nk hik q
m
ðc^ hf 2  v^ 2 Þ   ðc^ hfn  v^ n Þ; h; f ¼ 1; . . . ; q: ð13Þ and the collaborative utility matrix C~ jk ¼ ½c~ hfjk q
q .
Step 2: Represent w~ ik , v~ jk , d~ hik and c~ hfjk into triangular fuzzy
L M R L M R
If we note that c^ hf ¼ ðchf ; chf ; chf Þ, then chf , chf and chf are respectively numbers using Eq. (2), i.e., w ^ ik ¼ ðwLik ; wM R ^ L M R
ik ; wik Þ, v jk ¼ ðvjk ; vjk ; vjk Þ,
calculated by d^ hik ¼ ðdLhik ; dM R ^ L M R
hik ; dhik Þ and c hfjk ¼ ðchfjk ; chfjk ; chfjk Þ, respectively.
Pn L L
L j ¼ 1 vj chfj Step 3: Obtain the collective individual attribute weight vector
chf ¼ ; h; f ¼ 1; . . . ; q; ð14aÞ ^ ¼ ðw
n W ^ 2; . . . ; w
^ 1; w ^ m Þ and the collective individual utility matrix
Pn M M
^ ¼ ½d^ hi 
D using Eqs. (3)–(6).
q
m
M j ¼ 1 vj chfj
chf ¼ ; h; f ¼ 1; . . . ; q; ð14bÞ Step 4: Compute the overall assessment value of individual
n
Pn utility of candidate partner d^ using Eqs. (7) and (8).
h
R R
R j ¼ 1 vj chfj Step 5: Obtain the collective collaborative attribute weight
chf ¼ ; h; f ¼ 1; . . . ; q: ð14cÞ
n vector V^ ¼ ðv^ 1 ; v^ 2 ; . . . ; v^ n Þ and the collective collaborative utility
Let c^ h represent the overall assessment value of collaborative matrix C^ j ¼ ½c^ hfj q
q using Eqs. (9)–(12).
utility between candidate Ph and other candidates and it is Step 6: Obtain the overall collaborative utility matrix
expressed as
C^ ¼ ½c^ hf q
q using Eqs. (13) and (14).
c^ h ¼ ð1=qÞ  ½ch1  ch2   chq ; h ¼ 1; . . . ; q; ð15Þ Step 7: Compute the overall assessment value of collaborative
where c^ h is a triangular fuzzy number. If we note that utility of each candidate partner c^ h using Eqs. (15) and (16).
Step 8: Compute the overall assessment value of each partner
c^ h ¼ ðchL ; chM ; chU Þ, then there are
Pq j^ h using Eqs. (17) and (18), and get the corresponding crisp
f ¼ 1 hf
cL number jdef using Eq. (19).
chL ¼ ; h ¼ 1; . . . ; q; ð16aÞ h
q
Step 9: Rank the all candidate partners according to jdef
h
and
Pq select the desired partner(s) for a codevelopment alliance.
cM
f ¼ 1 hf
chM ¼ ; h ¼ 1; . . . ; q; ð16bÞ
q
Pq 6. Application of the proposed method
cR
f ¼ 1 hf
chR ¼ ; h ¼ 1; . . . ; q: ð16cÞ
q
Baosight is one of the top five software companies in China,
CMM/CMMi Level 5 accredited and it is amongst ‘‘The Ten Most
5.4. Integration of individual and collaborative utilities Innovative Software Companies in China’’ by China Software
Industry Association in 2007 and 2008. It owns 143 registered
To obtain the overall assessment value of each candidate software products, 150 patents and 261 registered software
partner, we integrate the overall assessment value of individual copyrights. Baosight provides its clients with all-around, whole
utility in Eq. (7) and that of collaborative utility in Eq. (15). lifecycle software development design service, spanning from
Suppose that the decision-maker gives the weights of design, development, coding, testing and maintenance to applica-
individual and collaborative utilities as a and b, such that tion development. IT software service of Baosight is featured
a þ b ¼ 1 and 0 r a; b r 1. The overall assessment value of mainly in the areas of Embedded Development, Business
candidate Ph can be obtained by Intelligence (BI), Collaborative Commerce, Manufacturing Execu-
tion System (MES) and Enterprise Resource Planning (ERP).
j^ h ¼ ad^ h þ bc^ h ; h ¼ 1; . . . ; q; ð17Þ Baosight takes codevelopment alliance as one of its important
where j ^ h ¼ ðlLh ; lM
^ h is a triangular fuzzy number, i.e., j U strategy to hold its core competence in NPD. From 1999, it allies
h ; lh Þ, then
with a number of firms and research institutions for developing new
lLh , lM R
h and lh are respectively calculated by products such as collaborative work system, MES in process
lLh ¼ adLh þ bchL ; h ¼ 1; . . . ; q; ð18aÞ industry, workflow-based system. Through alliances, Baosight aims
at: (i) reducing the NPD developing time, (ii) decreasing the NPD
lM M M
h ¼ adh þ bch ; h ¼ 1; . . . ; q; ð18bÞ cost, (iii) obtaining partners’ complementary core competence to fill
knowledge gap, and (iv) easily accessing to a new market.
lRh ¼ adRh þ bchR ; h ¼ 1; . . . ; q: ð18cÞ MES is complicated especially in steel firms that belong to the
process industry. Thus, two partners need to be selected from
Using Eq. (1), j
^ h is mapped into a crisp number to obtain the
seven candidate software companies to form a codevelopment
ranking value of each candidate partner, i.e.,
alliance for MES development. The basic information about the
D½ðlM R M 2 L R 2 M L 2
h  LÞðD þ lh  lh Þ ðR  lh Þþ ðlh  LÞ ðD þ lh  lh Þ  seven candidate partners is listed in Table 5.
jdef
h
¼ Lþ M L R M L R M L R M
;
ðD þ lh  lh ÞðD þ lh  lh Þ2 ðR  lh Þþ ðlh  LÞðD þ lh  lh Þ2 ðD þ lh  lh Þ Five experts were invited to conduct the partner selection task,
ð19Þ herein, two from Baosight (E1 and E2) and three from external
L L L R R R organizations (E3, E4 and E5). Each expert distributes in a certain
where L ¼ minfl1 ; l2 ; . . . ; lq g, R ¼ maxfl1 ; l2 ; . . . ; lq g, D ¼ R  L. field of NPD, technology, engineering, strategy management or
According to overall assessment value def
h
j, the ranking order of marketing. A senior manager of NPD department guides the
ARTICLE IN PRESS
166 B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170

Table 5 Table 8
The basic information about the seven candidate partners. The linguistic assessments of collaborative attribute ratings provided by expert E1.

No. Candidate Found time Number of 2006 sales P1 P2 P3 P4 P5 P6 P7


partners employee volume
C1
1 FW 1998 196 RMB 38.3 M P1 H M H VL L H DH
2 BJ 2003 78 RMB 8.9 M P2 M L DH M VL L VL
3 NSF 1990 10,021 RMB 1808 M P3 H DH VH H M VH M
4 XC 1997 342 RMB 34.4 M P4 VL DH H M DL M H
5 DEK 1993 1229 RMB 246.3 M P5 L VL M DL M H L
6 GURU 2000 2780 RMB 398.2 M P6 H L VH M H H VL
7 KG 1995 5812 RMB 1002 M P7 DH VL M H L VL VH

C2
P1 M L H M DH VL M
P2 L M M DH L H DL
P3 H M VH VH H L DL
P4 M DH VH M M VL H
Table 6 P5 DH L H M L VH M
The judgments of attribute weights provided by the five experts. P6 VL H L VL VH VH L
P7 M DL DL H M L VH
Experts Individual attributes Collaborative attributes
C3
I1 I2 I3 I4 C1 C2 C3 C4 C5 P1 H H M L DL H VL
P2 H L L DL M L VH
E1 DH VH VH H VH DH H VH VH P3 M L H H M VH DH
E2 H VH VH M DH DH H VH H P4 L DL H M VL H M
E3 DH H DH H DH DH VH VH H P5 DL M M VL VL DH L
E4 DH M VH DH DH DH H H VH P6 H L VH H DH M H
E5 DH M H M DH VH H H H P7 VL VH DH M L H H

C4
P1 H VL VH M DL H M
P2 VL M M VH VL H L
P3 VH M VH DH H VL VH
Table 7 P4 M VH DH L M L H
The linguistic assessments of individual attribute ratings provided by the five P5 DL VL H M M VL M
experts. P6 H H VL L VL VH VL
P7 M L VH H M VL M
Experts Candidate partners Individual attributes
C5
I1 I2 I3 I4 P1 L DH M L VH M DH
P2 DH H M VH DL H VL
E1 P1 M H M VH P3 M M M L DL M VH
P2 DL DH M VH P4 L VH L H VL H M
P3 DL M L VH P5 VH DL DL VL L DL DH
P4 DL VL M L P6 M H M H DL H VH
P5 H M DH M P7 DH VL VH M DH VH DH
P6 M L M DH
P7 DH VL H M

E2 P1 L VH VL M Table 9
P2 H M VL L The linguistic assessments of collaborative attribute ratings provided by expert E2.
P3 M L H M
P4 L H VL VH P1 P2 P3 P4 P5 P6 P7
P5 VH H VH L
P6 H VL DL M C1
P7 VH M VL H P1 M M H DH L M VL
P2 M L M L VL DH M
E3 P1 DL DH M H
P3 H M H L H M VH
P2 L DH H M
P4 DH L L M VH L M
P3 H VH M L
P5 L VL H VH L M DL
P4 H M L M
P6 M DH M L M H H
P5 H VH M VH
P7 VL M VH M DL H DH
P6 VL M H VL
P7 M L VH DH C2
P1 M L M VL H DH VL
E4 P1 M L H DH
P2 L VL H L DL VH M
P2 VH H M M
P3 M H H VL M VH L
P3 VL H VL H
P4 VL L VL M VL L DH
P4 M L DL M
P5 H DL M VL L H VL
P5 M DH L M
P6 DH VH VH L H H M
P6 H VH L M
P7 VL M L DH VL M M
P7 H DL DL M
C3
E5 P1 H M L M
P1 L VL M H DL M VH
P2 M L DL H
P2 VL L VH L M H VL
P3 L M DL L
P3 M VH DH DH VH L H
P4 VL M L H
P4 H L DH M M H M
P5 L VL H H
P5 DL M VH M L L VH
P6 L M DL VH
P6 M H L H L H DL
P7 L H VL H
P7 VH VL H M VH DL M
ARTICLE IN PRESS
B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170 167

Table 9 (continued ) Table 11


The linguistic assessments of collaborative attribute ratings provided by expert E4.
P1 P2 P3 P4 P5 P6 P7
P1 P2 P3 P4 P5 P6 P7

C4
C1
P1 H M H DL VL H VH
P1 L L DH M VL H L
P2 M M M VL H DL M
P2 L M VL H M DL VH
P3 H M M H L DH VH
P3 DH VL VH VH H VH L
P4 DL VL H H M VH DL
P4 M H VH H DL H VL
P5 VL H L M L L L
P5 VL M H DL VL M M
P6 H DL DH VH L M VL
P6 H DL VH H M H DH
P7 VH M VH DL L VL DH
P7 L VH L VL M DH DH
C5
C2
P1 M VL H M VL L M
P1 L H VH M DL M VL
P2 VL H M VH DL H VL
P2 H M L M VL H M
P3 H M H H M DH L
P3 VH L VL VH M VH DH
P4 M VH H L H DL VL
P4 M M VH H L L M
P5 VL DL M H M VL M
P5 DL VL M L VL VL L
P6 L H DH DL VL DH DH
P6 M H VH L VL H VH
P7 M VL L VL M DH H
P7 VL M DH M L VH DH

C3
Table 10 P1 M H VH M DL H L
The linguistic assessments of collaborative attribute ratings provided by expert E3. P2 H M H VL M DH L
P3 VH H L DH M L VH
P1 P2 P3 P4 P5 P6 P7 P4 M VL DH H VL M DH
P5 DL M M VL L DL M
C1 P6 H DH L M DL VH VL
P1 H M VH VL M L VL P7 L L VH DH M VL H
P2 M L DL M VH M DH
C4
P3 VH DL DH DH L VH M
P1 M L VH M L VL M
P4 VL M DH M VL M H
P2 L M H L DL H DH
P5 M VH L VL VL DL DL
P3 VH H M M H DH L
P6 L M VH M DL H H
P4 M L M M DL H VL
P7 VL DH M H DL H DH
P5 L DL H DL M L M
C2 P6 VL H DH H L VH L
P1 H DL H L VL VH M P7 M DH L VL M L VH
P2 DL M VH H L DH VL
C5
P3 H VH DH DH L H VL
P1 M L VH L M VL M
P4 L H DH M H M DH
P2 L L H VL H M VH
P5 VL L L H L VL M
P3 VH H M DH VL VH M
P6 VH DH H M VL VH L
P4 L VL DH H L H DL
P7 M VL VL DH M L VH
P5 M H VL L H DL L
C3 P6 VL M VH H DL M M
P1 M H M DL VL M H P7 M VH M DL L M M
P2 H M DH DL VH L VL
P3 M DH H L H M DH
P4 DL DL L H M VH M
P5 VL VH H M H H VH
P6 M L M VH H H L
P7 H VL DH M VH L VH

C4 Table 12
P1 L L DH M VL VH H The linguistic assessments of collaborative attribute ratings provided by expert E5.
P2 L M M H L L VL
P3 DH M H DH DL H VH P1 P2 P3 P4 P5 P6 P7
P4 M H DH M VH VL DL
P5 VL L DL VH M DL L C1
P6 VH L H VL DL VH M P1 M VL VH M DL H DL
P7 H VL VH DL L M VH P2 VL M M H H VL M
C5 P3 VH M DH DH VL VH L
P1 M H DH VL DL H VL P4 M H DH M VH L VH
P2 H L L M H DL M P5 DL H VL VH L VL H
P3 DH L M DH M M H P6 H VL VH L VL H DL
P4 VL M DH M VL H VH P7 DL M L VH H DL H
P5 DL H M VL L VL H C2
P6 H DL M H VL H DH P1 M M L DH VL H L
P7 VL M H VH H DH M P2 M M DH M H VL M
P3 L DH DH H VL M DH
P4 DH M H M M VH VL
P5 VL H VL M L H VH
partner selection task. He regards that the importance weights of P6 H VL M VH H H VL
the five experts are the same. To implement the partner selection P7 L M DH VL VH VL VH
task, first, he declares the strategy, mission, objective and related C3
requirements to the five experts clearly. Second, the attributes P1 M L H DL L VL M
(see Tables 2 and 3) for partner selection are explained to the five P2 L M DH VH VL DL L
experts in detail, in order that they could provide objective and P3 H DH DH VH M VH M
ARTICLE IN PRESS
168 B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170

Table 12 (continued ) precise responses. Last, the manager lets the five experts express
their opinions on the attribute weights and attribute ratings by
P1 P2 P3 P4 P5 P6 P7
questionnaires. The options of questions in questionnaires are the
P4 DL VH VH H H M H seven-ranking scale linguistic terms as defined in Section 5. The
P5 L VL M H L VL L expert selects one option from seven terms to answer every
P6 VL DL VH M VL H DH question. Table 6 shows their judgments on individual and
P7 M L M H L DH H collaborative attribute weights. Table 7 shows their linguistic
C4 assessments on individual attribute ratings. Tables 8–12 show
P1 L M DH VH M L DL their linguistic assessments on collaborative attribute ratings,
P2 M H DL M H VL DH
particularly, the diagonal elements are the assessments on
P3 DH DL H VL L H M
P4 VH M VL H DH M VL collaborative utility between Baosight and the candidate partners.
P5 M H L DH M M DL Using the approach proposed in Section 5, the overall assessment
P6 L VL H M M VH VH value of individual utility of each candidate partner is obtained using
P7 DL DH M VL DL VH VH Eqs. (3)–(8), which are shown in Table 13. The overall collaborative
C5 utility matrix is shown in Table 14 using Eqs. (9)–(16). Furthermore,
P1 H H DH L VL H DL the overall assessment value of collaborative utility of each
P2 H L M VL DL L H
candidate partner is obtained as follows:
P3 DH M H M VL DH M
P4 L VL M M VL H VH c^ 1 ¼ ð0:21; 0:39; 0:59Þ; c^ 2 ¼ ð0:20; 0:38; 0:58Þ; c^ 3 ¼ ð0:31; 0:52; 0:72Þ;
P5 VL DL VL VL DL DL L
P6 H L DH H DL M DH c^ 4 ¼ ð0:24; 0:43; 0:62Þ; c^ 5 ¼ ð0:15; 0:31; 0:50Þ; c^ 6 ¼ ð0:25; 0:45; 0:62Þ;
P7 DL H M VH L DH DH
c^ 7 ¼ ð0:25; 0:44; 0:63Þ:

Table 13
The assessment value of individual utility of each candidate partner.

Ph d^ h1 d^ h2 d^ h3 d^ h4 d^ h

P1 (0.27, 0.40, 0.57) (0.50, 0.67, 0.80) (0.27, 0.43, 0.60) (0.53, 0.70, 0.83) (022, 0.41, 0.62)
P2 (0.33, 0.47, 0.63) (0.53, 0.70, 0.80) (0.23, 0.37, 0.53) (0.40, 0.57, 0.73) (0.22, 0.40, 0.60)
P3 (0.20, 0.33, 0.50) (0.40, 0.57, 0.73) (0.20, 0.33, 0.50) (0.37, 0.53, 0.70) (0.17, 0.33, 0.53)
P4 (0.20, 0.33, 0.50) (0.27, 0.43, 0.60) (0.13, 0.27, 0.43) (0.40, 0.57, 0.73) (0.14, 0.30, 0.50)
P5 (0.43, 0.60, 0.77) (0.47, 0.63, 0.77) (0.50, 0.67, 0.80) (0.40, 0.57, 0.73) (0.27, 0.48, 0.68)
P6 (0.30, 0.47, 0.63) (0.30, 0.47, 0.63) (0.20, 0.30, 0.47) (0.43, 0.60, 0.73) (0.18, 0.35, 0.54)
P7 (0.50, 0.67, 0.80) (0.20, 0.33, 0.50) (0.23, 0.37, 0.53) (0.50, 0.67, 0.80) (0.22, 0.40, 0.59)

Table 14
The overall collaborative utility matrix.

P1 P2 P3 P4 P5 P6 P7

P1 (0.22, 0.42, 0.63) (0.18, 0.36, 0.56) (0.37, 0.60, 0.81) (0.18, 0.35, 0.54) (0.1, 0.23, 0.4) (0.25, 0.45, 0.66) (0.17, 0.34, 0.53)
P2 (0.18, 0.36, 0.56) (0.18, 0.36, 0.57) (0.27, 0.46, 0.66) (0.22, 0.41, 0.61) (0.16, 0.31, 0.51) (0.22, 0.39, 0.58) (0.21, 0.39, 0.58)
P3 (0.37, 0.6, 0.81) (0.27, 0.46, 0.66) (0.36, 0.59, 0.78) (0.37, 0.60, 0.78) (0.18, 0.36, 0.56) (0.35, 0.57, 0.78) (0.29, 0.49, 0.70)
P4 (0.18, 0.35, 0.54) (0.22, 0.41, 0.61) (0.37, 0.60, 0.78) (0.25, 0.45, 0.67) (0.18, 0.34, 0.54) (0.23, 0.42, 0.63) (0.24, 0.43, 0.61)
P5 (0.1, 0.23, 0.4) (0.16, 0.31, 0.51) (0.18, 0.36, 0.56) (0.18, 0.34, 0.54) (0.11, 0.28, 0.47) (0.14, 0.28, 0.46) (0.19, 0.36, 0.57)
P6 (0.25, 0.45, 0.66) (0.22, 0.39, 0.58) (0.35, 0.57, 0.78) (0.23, 0.42, 0.63) (0.14, 0.28, 0.46) (0.36, 0.59, 0.81) (0.24, 0.43, 0.45)
P7 (0.17, 0.33, 0.53) (0.21, 0.39, 0.58) (0.29, 0.49, 0.70) (0.24, 0.43, 0.59) (0.19, 0.36, 0.57) (0.24, 0.43, 0.42) (0.42, 0.66, 0.85)

Table 15
Sensitive analysis.

a b Overall assessment values Ranking order of all candidates

jdef
1 jdef
2 jdef
3 jdef
4 jdef
5 jdef
6 jdef
7

1 0 0.4088 0.3971 0.3598 0.3271 0.4511 0.3661 0.4026 P5 gP1 gP7 gP2 gP6 gP3 gP4
0.9 0.1 0.4078 0.3966 0.3767 0.3385 0.4372 0.3748 0.4069 P5 gP1 gP7 gP2 gP3 gP6 gP4
0.8 0.2 0.4068 0.3960 0.3937 0.3499 0.4233 0.3834 0.4111 P5 gP7 gP1 gP2 gP6 gP3 gP4
0.7 0.3 0.4058 0.3955 0.4106 0.3613 0.4094 0.3921 0.4154 P7 gP3 gP5 gP1 gP2 gP6 gP4
0.6 0.4 0.4058 0.3955 0.4106 0.3613 0.4094 0.3921 0.4154 P7 gP3 gP5 gP1 gP2 gP6 gP4
0.5 0.5 0.4048 0.3950 0.4275 0.3727 0.3955 0.4007 0.4197 P3 gP7 gP1 gP6 gP5 gP2 gP4
0.4 0.6 0.4038 0.3944 0.4444 0.3841 0.3816 0.4094 0.4240 P3 gP7 gP6 gP1 gP2 gP4 gP5
0.3 0.7 0.4029 0.3939 0.4614 0.3955 0.3677 0.4181 0.4283 P3 gP7 gP6 gP1 gP4 gP2 gP5
0.2 0.8 0.4019 0.3933 0.4783 0.4069 0.3537 0.4267 0.4326 P3 gP7 gP6 gP4 gP1 gP2 gP5
0.1 0.9 0.0147 0.0140 0.0158 0.0122 0.0136 0.0143 0.0158 P7 gP3 gP1 gP6 gP2 gP5 gP4
0 1 0.3999 0.3922 0.5121 0.4298 0.3259 0.4440 0.4412 P3 gP6 gP7 gP4 gP1 gP2 gP5
ARTICLE IN PRESS
B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170 169

Moreover, varying a and b are used to integrate the assess- Science Fund of China (Project nos. 70901027, 70871021 and
ment values of individual and collaborative utilities. The overall 90924016).
assessment value of each candidate partner is obtained using Eqs.
(17)–(19), as shown in Table 15. It obvious that the ranking order
References
of candidate partners varies with changing a and b. For example,
when a =0.8 and b =0.2, the ranking order of candidate partners is
Afonso, P., Nunes, M., Paisana, A., Braga, A., 2008. The influence of time-to-market
P5 gP7 gP1 gP2 gP3 gP6 gP4 . In this case, candidates P5, P7 and and target costing in the new product development success. International
P1 are the desired partners. When a = 0.2 and b =0.8, the ranking Journal of Production Economics 115 (2), 559–568.
order of candidate partners changes into P3 gP7 gP6 g Amid, A., Ghodsypour, S.H., Brien, C.O., 2006. Fuzzy multiobjective linear model for
supplier selection in a supply chain. International Journal of Production
P4 gP1 gP2 gP5 , candidates P3, P7 and P6 are then the Economics 104 (2), 394–407.
appropriate ones. The sensitive analysis can support the Araz, C., Ozkarahan, I., 2007. Supplier evaluation and management system for
decision-maker to select the desired partners complying with strategic sourcing based on a new multicriteria sorting procedure. Interna-
tional Journal of Production Economics 106 (2), 585–606.
the company’s strategy goal in which the individual utility or the
Arino, A., 2003. Measures of strategic alliance performance: an analysis of
collaborative utility is focused on. construct validity. Journal of International Business Studies 34 (1), 66–79.
Bronder, C., Pritzl, R., 1992. Developing strategic alliances: a conceptual framework
for successful co-operation. European Management Journal 10 (4), 412–422.
Büyüközkan, G., Feyzioglu, O., Nebol, E., 2008. Selection of the strategic alliance
7. Conclusions ˘
partner in logistics value chain. International Journal of Production Economics
113 (1), 148–158.
This paper proposes a method for partner selection of Chang, S.L., Wang, R.C., Wang, S.Y., 2006. Applying fuzzy linguistic quantifier to
select supply chain partners at different phases of product life cycle.
codevelopment alliances using individual and collaborative International Journal of Production Economics 100 (2), 348–359.
utilities. A framework is presented, which describes the individual Chen, S.M., 1996. A new method for evaluating weapon systems using fuzzy set
attributes and collaborative ones for partner selection, as well as theory. IEEE Transactions on Systems, Man and Cybernetics 26 (4), 493–497.
Chesbrough, H.W., 2003. Open Innovation: The New Imperative for Creating and
the distinct measurements of two types of attributes. A FMADM Profiting from Technology. Harvard Business School Press, Boston.
approach using individual and collaborative utilities is then Costantino, F., Gravio, G.D., 2009. Multistage bilateral bargaining model with
proposed to select the appropriate partners for codevelopment incomplete information—a fuzzy approach. International Journal of Produc-
tion Economics 117 (2), 235–243.
alliances. Additionally, an example is used to illustrate the
Cowan, R., Jonard, N., Zimmermann, J.B., 2007. Bilateral collaboration and the
potential applicability of the proposed method. The method emergence of innovation networks. Management Science 53 (7), 1051–1067.
proposed in this paper is convenient for processing linguistic Dacin, T., Hitt, M.A., Levitas, E., 1997. Selecting partners for successful international
alliances: examination of US and Korean firms. Journal of World Business 32
terms in a fuzzy environment. It is pragmatic for firms to conduct
(1), 3–16.
partner selection in the formation of a codevelopment alliance. Deck, M., Strom, M., 2002. Model of co-development emerges. Research
Several distinguished contributions of this study are as follows. Technology Management 45 (3), 47–54.
First, a framework for partner selection of codevelopment Ding, J.F., Liang, G.S., 2005. Using fuzzy MCDM to select partners of strategic
alliances for liner shipping. Information Sciences 173 (1–3), 197–225.
alliances is proposed. In the framework, a novel idea is presented Emden, Z., Calantone, R.J., Droge, C., 2006. Collaborating for new product
that not only the individual utility of each candidate partner, but development: selecting the partner with maximum potential to create value.
also collaborative utility between pairwise partners should be Journal of Product Innovation Management 23 (4), 330–341.
Famuyiwa, O., Monplaisir, L., Nepal, B., 2008. An integrated fuzzy-goal-program-
considered in partner selection. According to the framework, ming-based framework for selecting suppliers in strategic alliance formation.
straightforward and routine decision-making approaches can be International Journal of Production Economics 113 (2), 862–875.
developed to solve other practical partner selection problems. Fan, Z.P., Feng, B., Suo, W.L., 2009. A fuzzy linguistic method for evaluating
collaboration satisfaction of NPD team using mutual-evaluation information.
Second, an evaluation hierarchy, including individual and International Journal of Production Economics 122 (2), 547–557.
collaborative attributes, for partner selection of codevelopment Feng, B., Jiang, Z.Z., Fan, Z.P., Na, F., 2009. A method for member selection of Cross-
alliances is constructed. It provides the decision-maker with a Functional Teams using the individual and collaborative information.
European Journal of Operational Research, in press, Available online:
comprehensive description of individual and collaborative per-
/http://www.sciencedirect.comS.
formance of candidate partners and it lays a good foundation for Fischer, M., Jahn, H., Teich, T., 2004. Optimizing the selection of partners in
decision-making. production networks. Robotics and Computer-Integrated Manufacturing 20
(6), 593–601.
Third, a FMADM approach is proposed to select the desired
Geringer, J.M., 1988. Joint Venture Partner Selection: Strategies for Developed
partner for codevelopment alliances. The approach is suitable to Countries. Quorum Books, Westport.
process and aggregate the assessment data that are in the form of Hajidimitriou, Y.A., Georgiou, A.C., 2002. A goal programming model for partner
linguistic terms. Moreover, it provides a formal procedure to selection decisions in international joint ventures. European Journal of
Operational Research 138 (3), 649–662.
integrate the individual utility of each candidate partner and Huang, X.G., Wong, Y.S., Wang, J.G., 2004. A two-stage manufacturing partner
collaborative utility between pairwise partners. Using the pro- selection framework for virtual enterprises. International Journal of Computer
posed approach, not only individual utility but also collaborative Integrated Manufacturing 17 (4), 294–304.
Kaufmann, A., Gupta, M.M., 1985. Introduction to Fuzzy Arithmetic: Theory and
utility is involved in the ranking result of candidates. Applications. Thomson Computer Press, Van Nostrand Reinhold, New York.
Moreover, it should be pointed out that partner selection is a Li, D.F., Yang, J.B., 2004. Fuzzy linear programming technique for multiattribute
complicated problem in the real world. Besides linguistic terms group decision making in fuzzy environments. Information Sciences 158,
263–275.
used in this paper, the data in other forms may be involved at the Lin, C.J., Wu, W.W., 2008. A causal analytical method for group decision-making
same time. Therefore, straightforward and pragmatic decision- under fuzzy environment. Expert Systems with Applications 34 (1), 205–213.
making approaches need to be investigated with regard to the Mikhailov, L., 2002. Fuzzy analytical approach to partnership selection in
formation of virtual enterprises. Omega 30 (5), 393–401.
real-world situations.
Opricovic, S., Tzeng, G.H., 2003. Defuzzification within a multicriteria decision
model. Journal of Uncertainty, Fuzziness and Knowledge Based Systems 11 (5),
635–652.
Acknowledgements Perks, H., 2000. Marketing information exchange mechanisms in collaborative new
product development. Industrial Marketing Management 29 (2), 179–189.
Saen, R.F., 2007. Suppliers selection in the presence of both cardinal and ordinal
This work was partly supported by the National Natural data. European Journal of Operational Research 183 (2), 741–747.
Science Fund for Distinguished Young Scientists of China (Project van Laarhoven, P.J.M., Pedrycz, W., 1983. A fuzzy extension of Saaty’s priority
theory. Fuzzy Sets and Systems 11 (1–3), 229–241.
no. 70525002), National Natural Science Fund for Creative Verma, R., Pullman, M.E., 1998. An analysis of the supplier selection process.
Research Groups of China (Project no. 70721001), and National Omega 26 (6), 739–750.
ARTICLE IN PRESS
170 B. Feng et al. / Int. J. Production Economics 124 (2010) 159–170

Vonderembse, M.A., Tracey, M., 1999. The impact of supplier selection criteria and Zadeh, L.A., 1975. The concept of a linguistic variable and its application to
supplier involvement on manufacturing performance. Journal of Supply Chain approximate reasoning. Information Science 8 (3), 199–249 (I) 8 (4), 301–357
Management 35 (3), 33–39. (II).
Wang, T.C., Chen, Y.H., 2007. Applying consistent fuzzy preference relations to Zeng, Z.B., Li, Y., Zhu, W.X., 2006. Partner selection with a due date constraint in
partnership selection. Omega 35 (4), 384–388. virtual enterprises. Applied Mathematics and Computation 175 (2),
Zadeh, L.A., 1965. Fuzzy sets. Information and Control 8, 338–353. 1353–1365.

You might also like