Asia Pacific Management Review 16(4) (2011) 491-502

www.apmr.management.ncku.edu.tw

Applying Linguistic VIKOR and Knowledge Map in Personnel Selection

Chen-Tung Chen a,, Ping-Feng Pai b, Wei-Zhan Hung c
a
Department of Information Management, National United University, Taiwan
b
Department of Information Management, National Chi Nan University, Taiwan
c
Department of International Business Studies, National Chi Nan University, Taiwan
Received 21 May 2010; Received in revised form 18 October 2011; Accepted 9 November 2011

Abstract

In the high competitive environment, both core knowledge of business and intelligent
employee are important assets of a company. Personnel selection is a very important issue for
the enterprises because most of the enterprises’ activities need the right person to handle it. In
order to choose the suitable employee for the company to compete with other competitors, it
needs a suitable method to deal with the problem of personnel selection based on the
knowledge map of company and capability of each employee. In fact, many factors will
influence the personnel selection problem such as language ability, work experience,
communication ability, and so on. By considering general criteria and knowledge criteria
simultaneously, a linguistic VIKOR method is proposed in this paper to deal with personnel
selection problem. Finally, an example is implemented to demonstrate the practicability of
the proposed method.

Keywords: Knowledge map, fuzzy set, personnel selection, MCDM

1. Introduction

In the knowledge economy age, a knowledgeable and talented employee is the company’s
asset (Wei, 2007). Personnel selection is a very important issue for the enterprises because
most of the enterprises’ activities need the right person to handle it (Caligiuri et al., 2009). On
the contrary, the person in the wrong position will reduce the company’s competitiveness.
Moreover, Stein and Zwass (1995) stated that enterprise can aggregate organization memory
(Intelligent Capital) to create competitiveness. So, knowledge had been considered as a
crucial asset for companies to survive in a highly competitive environment (Lai et al., 2008).
It is important for identifying, disseminating, categorizing, retrieving and sharing information
assets (knowledge) throughout the organization (Taft, 2000). There are two kinds of
knowledge such as explicit and tacit. Explicit knowledge refers to codified knowledge that is
transmittable in formal, systematic language and is easily transferred by using information
technology (IT) (Woo et al., 2004). Tacit knowledge is knowledge housed in the human brain,
such as expertise, understanding, or professional insight formed as a result of experience
(Polanyi, 1996). However, most of the competitive knowledge is tacit knowledge.
Knowledge map is a tool or technique for visualizing knowledge and relationship that the
relevant features of the knowledge can be clearly highlighted (Vail, 1999). The function of
knowledge map is to record the person who possess the special knowledge in the company.
So, knowledge map is also called “yellow pages”. The goal of knowledge map is to make
sure that manager or relative workers can find employee who possess the relative knowledge
in special case (Earl, 2001). Hence, manager can save the resource and reduce the operation
time to record tacit knowledge which is difficult to express clearly. According to the


Corresponding author. Email: ctchen@nuu.edu.tw

491
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

literatures, knowledge map has been used in many situations. Woo et al., (2004) used
software (such as Dynamic Knowledge Map) to improve the job performance in Architecture,
Engineering and Construction (AEC) company. Pyo (2005) compared four tourism
destination types of knowledge maps in tourism industry and suggested different mapping
schemes to let traveler find useful travel information. Rouse et al., (1998) developed a six-
step knowledge mining methodology according to user-oriented knowledge maps to judge
knowledge available in a R&D project. Shaw (2010) proved that the application of
knowledge maps in e-learning materials design can improve workers’ learning performance.
It is essential to choose employee which can provide the relative knowledge and experience
based on company’s knowledge requirement in accordance with company’s knowledge map.
There are some literatures about dealing with personnel selection problem. Ertugrul
Karsak (2000) used fuzzy objective programming to select personnel. Korvin et al., (2002)
used the fuzzy construct of compatibility to measure the fitness degree of a person’s skill set.
It combined levels of compatibility with acceptable levels of quality to deal with multiple
phase project personnel selection problem. Chien and Chen (2007) developed an effective
data mining approach based on rough set theory to analyze human resource data for personnel
selection. Chien and Chen (2008) developed a data mining framework based on decision tree
and association rules to generate useful rules for personnel selection. Gungor et al., (2009)
used fuzzy AHP to cope with personnel selection problem by considering both quantitative
and qualitative criteria. Celik et al., (2009) used fuzzy AHP and Fuzzy TOPSIS to recruit
academic personnel in maritime human resources. Fan et al., (2009) used a bi-objective 0-1
programming model to select desired members by considering individual information of
members and the collaborative information between members. Feng et al., (2010) built a
multi-objective 0–1 programming model and improved non-dominated sorting genetic
algorithm II (INSGA-II) based on the individual and collaborative performances of candidate
to deal with the problem of project member selection. Although the knowledge map is an
important tool for selecting personnel, a few of above literatures considered about this issue.
Furthermore, there are several conflicting criteria (such as experience and age) need to be
considered in choosing the suitable employee. Therefore, multi-criteria decision making
method is suitable to deal with the personnel selection problems.
In multi-criteria analysis, it is difficult to obtain a solution to satisfy all criteria
simultaneously (Chang and Hsu, 2009). A compromise solution for a problem with
conflicting criteria can help decision-makers identify an acceptable answer (Opricovic, 1998).
VIKOR (VlseKriterijumska Optimizacija I Kompromisno Resenje) is a multi criteria decision
making method which developed by Opricovic (1998). VIKOR determines a compromise
solution that provides the maximum group utility for the majority and a minimum of
individual regret for the opponent (Chen and Wang, 2009). So, VIKOR can find a
compromise priority ranking of alternatives according to the evaluation criteria (Opricovic
and Tzeng, 2002, 2003). According to the VIKOR method, the compromise ranking order
could be performed by comparing the measure of closeness to the ideal alternative through
the process of ranking and selecting a set of alternatives in the presence of conflicting criteria
(Chen and Wang, 2009). Because imprecise and subjective information that often appears in
personnel evaluation process, crisp values are inadequate for solving the problems. A more
realistic approach may be to use linguistic assessments instead of numerical values (Chen,
2000; Herrera and Herrera-Viedma, 2000). The 2-tuple linguistic representation model is
based on the concept of symbolic translation (Herrera and Martinez, 2000; Xu, 2005).
Decision makers can apply 2-tuple linguistic variables to express their opinions and obtain
the final evaluation result with appropriate linguistic variable. It is an effective method to
reduce the mistakes of information translation and avoid information loss through computing
with words (Herrera-Viedma, 2003). Therefore, a linguistic VIKOR method is proposed to

492
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

evaluate employee by both considering general criteria and knowledge criteria based on
knowledge map in the company. Linguistic VIKOR method not only let experts flexibly use
different kind of linguistic variables to express their opinions but also obtain a compromise
solution which take account of the maximum group utility for the majority and a minimum of
individual regret for the opponent simultaneously.
This paper is organized as follows. In section 2, we present the definition and operation of
2-tuple linguistic variables. In section 3, we describe the detail of the proposed method. In
section 4, an example is implemented to demonstrate the procedure for the proposed method.
Finally, the conclusion is discussed at the end of this paper.

2. 2-Tuple Linguistic Variable

Let S  {s 0 , s1 , s 2 ,..., s g } be a finite and totally ordered linguistic term set. A 2-tuple
linguistic variable can be expressed as ( s i ,  i ) , where si is the central value of i-th
linguistic term in S and  i is a numerical value representing the difference between
calculated linguistic term and the closest index label in the initial linguistic term set. The
symbolic translation function  is used to translate β into a 2-tuple linguistic variable
(Herrera and Martinez, 2001). Then, the symbolic translation process is applied to translate β
(β [0, 1]) into a 2-tuple linguistic variable. The generalized translation function can be
represented as (Tai and Chen, 2009):
1 1 i
 : [0,1]  S  [ , ) and (  )  ( si ,  i ) where i  round (   g ) ,  i    and
2g 2g g
1 1
 i  [ , ).
2g 2g
A reverse function 1 is defined to return an equivalent numerical value β from 2-tuple
linguistic information ( s i ,  i ) . According to the symbolic translation, an equivalent
numerical value β is obtained as follow (Tai and Chen, 2009).
i
1( si , i )    i   (1)
g
Let x = {(r1, 1), (r2, 2)…, (rn, n)} be a 2-tuple linguistic variable set. The arithmetic
mean X is computed as (Herrera-Viedma et al., 2004)
1 n 
X    1(ri , i )   ( sm , m ) (2)
 n i 1 
In general, decision makers would use the different 2-tuple linguistic variables based on
their knowledge or experiences to express their opinions (Herrera et al., 2005). A
transformation function is needed to transfer these 2-tuple linguistic variables from different
linguistic sets to a standard linguistic set at unique domain. However, the domain of the
linguistic variables will increase as the number of linguistic variable is increased in the
method of Herrera and Martinez (Goumas and Lygerou, 2000). To overcome this drawback, a
new translation function is proposed here to transfer a crisp number or 2-tuple linguistic
variable to a standard linguistic term at the unique domain (Tai and Chen, 2009). Suppose
that the interval [0, 1] is the unique domain. The linguistic variable sets with different
semantics (or types) will be defined by partitioning the interval [0, 1]. Transforming a crisp
number β (β [0, 1]) into i-th linguistic term ( sin ( t ) , in ( t ) ) of type t as t (  )  ( sin ( t ) ,  in ( t ) ) where
i
i  round (   gt ) ,  in ( t )    g t  n(t )  1 and n (t) is the number of linguistic variable of type t.
gt

493
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

Transforming i-th linguistic term of type t into a crisp number β (β [0, 1]) as
i
t 1 ( sin(t ) ,  in(t ) )    in(t )   (3)
gt
1 1
where gt  n(t )  1 and  in ( t )  [ , ).
2 gt 2 gt
Therefore, the transformation from i-th linguistic term ( sin ( t ) , in ( t ) ) of type t to k-th
linguistic term ( skn ( t 1) , kn ( t 1) ) of type t+1 at interval [0, 1] can be expressed as
t 1(t 1( sin(t ) ,  in(t ) ))  ( skn(t 1) ,  kn(t 1) ) (4)
where gt 1  n(t  1)  1 and  n(t 1)  [ 1
,
1
).
k 2 g t 1 2 g t 1

3. Proposed Method

In fact, a personnel selection problem may be described by means of the following sets:
(a) A set of decision-makers is called D  D1 , D 2 ,  , D K  ;
(b) A set of candidates (alternatives) is called A  A1, A 2 ,  , A m  ;
(c) A set of general criteria C  C1 , C 2 ,  , C n  with which candidates’ performances are
measured;
(d) A set of performance ratings of candidates with respect to general criteria is called xij ,
i=1,2,…,m , j=1,2,…,n.

(e) A set of knowledge criteria F  F1, F2 ,, Fy based on company’s knowledge map 
with which candidates’ professional knowledge are measured;
(f) A set of professional ratings of candidates with respect to knowledge criteria is called
zid , i=1,2,…,m, d=1,2,…,y.
The performance of the i-th candidate with respect to the j-th general criterion decided by
the k-th expert can be represented as a 2-tuple linguistic variable Fjk  Ai   ( Sijk ,  ijk ) .The
aggregated linguistic ratings Fj  Ai  of the i-th alternative with respect to the j-th general
criterion can be calculated as
1 K 1 k
F j  Ai    ( k
  ( S ij ,  ij ))  ( S ij ,  ij ) (5)
K k 1
The aggregated linguistic weights ( w j ) of each general criterion can be calculated as
K
~  (1 
w 1 ( S wjk ,  wjk ))  ( S wj ,  wj )
j (6)
K k 1
The linguistic positive-ideal solution ( F * ) of each general criterion can be calculated as
j


 
F *j    Max   1 F j  Ai   ,  Ai  A  (7)
 i 
The linguistic negative-ideal solution ( F  ) of each general criterion can be calculated as
j


 
F j    Min   1 F j  Ai   ,  Ai  A  (8)
 i 
The group utility for the majority ( S i ) of each candidate can be calculated as
~ *  F    F  A 
1 * 1

  1 w
n
Si  j 1
j
j

 F    F 
1 *
, 1
j

i
i (9)
j j

494
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

where S i represents as the relative competitive ability of i-th candidate by weighted

average the relative position   F     F  A   of each general criterion.
1 * 1
j j i

 F    F 
1
j
* 1
j


The individual regret rating for the opponent R i of each candidate can be calculated as
 ~  *  F j    F j  A i   ,
1 * 1

R i  Max    1 w i (10)
  1 F j*     1 F j  
j
j

where R i represents as the maximum regret by choosing i-th candidate as solution
according to choose the worst performance in general criteria.
The value ( Q i ) of each candidate can be calculated as
Si  S * R  R*
Qi  v *  (1  v ) * i , i (11)
S S
 *
R  R*
S *  Min S i S   Max S i R *  Max R i R   Min R i (12)
,
i i , i , i

where S * represents the positive-ideal solution of S i , S  represents the negative-ideal

solution of S i , R * represents the positive-ideal solution of R i , R  represents the negative-
ideal solution of R i . The v value represents the decision making coefficient and v is between 0
and 1. When v is closed to 1 represents that decision maker chooses the candidate who
mainly considers maximize group utility for the majority. On the other hand, when v is closed
to 0 represents that decision maker chooses the candidate who mainly considers minimize
individual regret for the opponent. The smaller Q i represents i-th candidate is better with
respect to general criteria.
In this paper, the performances of candidates’ knowledge are measured according to
professional ratings of candidates with respect to knowledge criteria based on knowledge
map. The weight of knowledge criteria is according to the scarceness in the company. The
1
weight vector can be represented as Fw  Fw1 , Fw 2 ,, Fw y  where Fw d  Fd  F ,
FN d  1
FN d is the number of personnel in the company who possess the knowledge in knowledge
criteria Fd .
The zid is professional rating of i-th candidate with respect to knowledge criteria d. The
zid can be represented by 2-tuple linguistic variable. For example, the Fdk  Ai   ( Sidk ,  idk ) can
be represented as the opinion of k-th expert. The aggregated linguistic rating Fd  Ai  of the i-
th alternative with respect to the d-th knowledge criterion can be calculated as
1 K 1 k
F d  Ai    ( k
  ( S id ,  id ))  ( S id ,  id ) (13)
K k 1
We determine the threshold Ft  Ft1 , Ft 2 ,, Ft y  in each knowledge criterion to judge
that i-th candidate possesses the knowledge. The knowledge rating K id of i-th candidate with
respect to the d-th knowledge criterion can be calculated as
1, 1 Fd  Ai   Ft d
K id   (14)
0,  Fd  Ai   Ft d
1

If Fd  Ai  is equal to or larger than threshold Ft d , than we set K id to 1 and represent i-th

candidate possesses the knowledge in d-th knowledge criterion. Otherwise, we set Kid to 0.
The performance of knowledge PKi of each candidate can be calculated as

495
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

y
PK i   Fwd * K id (15)
d 1
The PKi represents the weighted average of the knowledge performance of i-th candidate.
The relative knowledge RK i of each candidate can be calculated as
maxPK i   PK i
RK i  (16)
maxPK i   min PK i 
The RK i value represents the knowledge competitiveness of each candidate. RK i is
between 0 and 1. The higher PK i , the lower RK i represents i-th candidate is more closed to
ideal-candidate with respect to knowledge criteria.
Considering general criteria and knowledge criteria simultaneously, the value RQ i of
each candidate can be calculated as
RQ i  u * Q i  (1  u ) * RK i ,  i , A i  A (17)
The u value represents the decision making coefficient and u is between 0 and 1. When u
is closed to 1 represents that decision maker chooses the candidate who mainly considers the
competitive ability of each candidate according to general criteria. On the other hand, when u
is closed to 0 represents that decision maker chooses the candidate who mainly considers
knowledge know-how of each candidate according to knowledge criteria based on company’s
knowledge map. According to the equation (17), the smaller RQ i represents i-th candidate is
better than other candidates with respect to all criteria.

4. Numerical Example

Suppose that a semi-conduct industry company desires to select an engineer. In the

company, three experts are formed a evaluation team to choose the best engineer from four
candidates in accordance with four general criteria and six knowledge expertise criteria based
on knowledge map. General criteria include English ability, work experience, communication
ability and emotional steadiness. Professional knowledge criteria include Japanese language
ability, product knowledge, market knowledge, accounting knowledge, law knowledge and
mechanism knowledge.
According to the proposed method, the computational procedures of the problem are
summarized as follows.
Step 1. Each expert will choose the linguistic variable type to express his opinion. Expert
D1 chooses type 1, D 2 chooses type 2, and D 3 chooses type 3 (refer to Table 1). Then, each
expert uses the linguistic variables to evaluate weights of each general criterion as Table 2
and performance ratings of each candidate with respect to general criteria as Table 3.

496
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

Table 1. Different types of linguistic variables.

Type Linguistic variable Figure
1 performance Extremely Poor ( s05 ) , Poor ( s15 ) , Fair ( s25 ) , Good ( s35 ) , Fig. 1
Extremely Good ( s45 )
weight Extremely Low ( s05 ) , Low ( s15 ) , Fair ( s25 ) , High ( s35 ) ,
Extremely High ( s45 )
2 performance Extremely Poor ( s07 ) , Poor ( s17 ) , Medium Poor ( s27 ) , Fig.2
Fair ( s37 ) , Medium Good ( s47 ) , Good ( s57 ) , Extremely
Good ( s67 )
weight Extremely Low ( s07 ) , Low ( s17 ) , Medium Low ( s27 ) , Fair ( s37 ) ,
Medium High ( s47 ) , High ( s57 ) , Extremely High ( s67 )
3 performance Extremely Poor ( s09 ) ,Very Poor ( s19 ) , Poor ( s 29 ) , Medium Fig. 3
Poor ( s39 ) , Fair ( s 49 ) , Medium Good ( s59 ) , Good ( s 69 ) , Very
Good ( s79 ) , Extremely Good ( s89 )
weight Extremely Low ( s09 ) ,Very Low ( s19 ) , Low ( s 29 ) , Medium
Low ( s39 ) , Fair ( s 49 ) , Medium High ( s59 ) , High ( s 69 ) , Very
High ( s79 ) , Extremely High ( s89 )

( s05 ) ( s15 ) ( s25 ) ( s35 ) ( s45 )

0 1

Figure 1. Membership functions of linguistic variables at type 1 (t=1).

( s 07 ) ( s17 ) ( s 27 ) ( s 37 ) ( s47 ) ( s57 ) ( s67 )

1
0

Figure 2. Membership functions of linguistic variables at type 2 (t=2).

497
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

( s09 ) ( s19 ) ( s29 ) ( s39 ) ( s49 ) ( s59 ) ( s69 ) ( s79 ) ( s89 )

0 1

Figure 3. Membership functions of linguistic variables at type 3 (t=3).

Table 2. The weight linguistic variable with respect to general criteria.

C1 C2 C3 C4
D1 ( s35 ,0) ( s45 ,0) ( s25 ,0) ( s25 ,0)
D2 ( s67 ,0) ( s57 ,0) ( s47 ,0) ( s57 ,0)
D3 ( s69 ,0) ( s79 ,0) ( s69 ,0) ( s79 ,0)

Table 3. The rating linguistic variable with respect to general criteria.

Criterion Alternative D1 D2 D3 Criterion Alternative D1 D2 D3
A1 ( s35 ,0) ( s67 ,0) ( s49 ,0) A1 ( s25 ,0) ( s57 ,0) ( s89 ,0)
A2 ( s45 ,0) ( s47 ,0) ( s39 ,0) A2 ( s35 ,0) ( s67 ,0) ( s79 ,0)
C1 C3
A3 ( s45 ,0) ( s47 ,0) ( s49 ,0) A3 ( s45 ,0) ( s57 ,0) ( s49 ,0)
A4 ( s25 ,0) ( s37 ,0) ( s59 ,0) A4 ( s45 ,0) ( s47 ,0) ( s89 ,0)
A1 ( s45 ,0) ( s57 ,0) ( s69 ,0) A1 ( s35 ,0) ( s57 ,0) ( s29 ,0)
A2 ( s35 ,0) ( s67 ,0) ( s29 ,0) A2 ( s15 ,0) ( s57 ,0) ( s29 ,0)
C2 C4
A3 ( s25 ,0) ( s37 ,0) ( s49 ,0) A3 ( s45 ,0) ( s37 ,0) ( s69 ,0)
A4 ( s45 ,0) ( s67 ,0) ( s49 ,0) A4 ( s15 ,0) ( s37 ,0) ( s59 ,0)

Step 2. Transform the linguistic evaluations of weight of each general criterion into the
linguistic variables of type 2 and aggregate the linguistic weight of each general criterion.
Step 3. Transform the linguistic ratings into the linguistic variables of type 2 and
aggregate the linguistic ratings of each candidate with respect to general criteria.
Step 4. Determine the linguistic positive-ideal solution F * and the linguistic negative-
j
ideal solution F  of each general criterion as Table 4.
j
Step 5. Compute the group utility for the majority S i , the individual regret rating for the
opponent R i and Q i when we set decision making coefficient v=0.5 as Table 5.
Step 6. Each expert uses the linguistic variables to evaluate the performance ratings of
each knowledge criterion as Table 6.
Step 7. Transform the linguistic ratings into the linguistic variables of type 2 and
aggregate the linguistic ratings of each candidate with respect to knowledge criteria.

498
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

Step 8. Collect data about the number of personnel who possess knowledge in each
knowledge criterion as Table 7.
Step 9. Determine the threshold values of each knowledge criterion as Table 7.
Step 10. Calculate the performance of knowledge PK i , the relative knowledge RK i and
RQ i when we set decision making coefficient u=0.5 as Table 8. Finally, the ranking order of
all candidates according to RQ i is A1  A2  A4  A3 .

Table 4. The values F * and F  of general criteria.

j j
C1 C2 C3 C4
F *j ( s37 ,0.063) ( s47 ,0.021) ( s57 ,0.056) ( s57 ,0.083)

F j ( s27 ,0.073) ( s27 ,0.042) ( s57 ,0.056) ( s37 ,0.056)

Table 5. The values S i , R i and Q i of each candidate.

A1 A2 A3 A4
Si 0.973 1.580 1.653 1.605
Ri 0.639 0.736 0.903 0.833
Qi 0.000 0.631 1.000 0.834

Table 6. The rating of linguistic variables of each expert with respect to knowledge
criteria.
Criterion Alternative D1 D2 D3 Criterion Alternative D1 D2 D3
A1 ( s15 ,0) ( s47 ,0) ( s49 ,0) A1 ( s45 ,0) ( s57 ,0) ( s39 ,0)
A2 ( s35 ,0) ( s67 ,0) ( s79 ,0) A2 ( s35 ,0) ( s37 ,0) ( s89 ,0)
F1 F4
A3 ( s25 ,0) ( s37 ,0) ( s79 ,0) A3 ( s15 ,0) ( s67 ,0) ( s39 ,0)
A4 ( s45 ,0) ( s37 ,0) ( s39 ,0) A4 ( s45 ,0) ( s37 ,0) ( s79 ,0)
A1 ( s25 ,0) ( s67 ,0) ( s69 ,0) A1 ( s25 ,0) ( s57 ,0) ( s79 ,0)
A2 ( s25 ,0) ( s47 ,0) ( s89 ,0) A2 ( s25 ,0) ( s37 ,0) ( s39 ,0)
F2 F5
A3 ( s25 ,0) ( s57 ,0) ( s59 ,0) A3 ( s15 ,0) ( s47 ,0) ( s49 ,0)
A4 ( s35 ,0) ( s67 ,0) ( s59 ,0) A4 ( s35 ,0) ( s67 ,0) ( s69 ,0)
A1 ( s15 ,0) ( s67 ,0) ( s89 ,0) A1 ( s45 ,0) ( s37 ,0) ( s89 ,0)
A2 ( s45 ,0) ( s37 ,0) ( s69 ,0) A2 ( s25 ,0) ( s67 ,0) ( s39 ,0)
F3 F6
A3 ( s35 ,0) ( s37 ,0) ( s39 ,0) A3 ( s35 ,0) ( s67 ,0) ( s49 ,0)
A4 ( s25 ,0) ( s57 ,0) ( s89 ,0) A4 ( s35 ,0) ( s37 ,0) ( s39 ,0)

499
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

Table 7. The number and threshold values of knowledge criteria.

F1 F2 F3 F4 F5 F6
Number 11 8 4 7 14 4
Threshold 0.667 0.667 0.667 0.667 0.667 0.667

Table 8. The values PK i , RKi and RQ i of each candidate.

A1 A2 A3 A4
PK i 0.703 0.519 0.200 0.503
RKi 0.000 0.365 1.000 0.398
RQ i 0.000 0.498 1.000 0.616

5. Conclusion

In the proposed method, experts can flexibly use their preferable linguistic variable to
express their opinions about the performance of each candidate which includes general
criteria and knowledge criteria. General criteria can be used to evaluate the competitive
ability of each candidate. Knowledge criteria are according to company’s knowledge map
and considering the scarceness of the knowledge. The linguistic VIKOR method not only can
deal with conflicting criteria but also can determine a compromise solution that considers the
maximum group utility for the majority and a minimum of individual regret for the opponent
simultaneously. The personnel selection problem is suitable to be dealt with by the linguistic
VIKOR method because it includes some conflicting criteria and needs to consider the
relative competitiveness of each candidate. In fact, we will combine knowledge map with
new methodology such as social network to improve the linguistic VIKOR method as our
next research topics. A computerized system will be designed based on linguistic VIKOR
method to deal with personnel promotion selection and other decision-making problems.

References
Caligiuri, P., Tarique, I., Jacobs, R. (2009) Selection for international assignments. Human
Resource Management Review, 19(3), 251-262.
Celik, M., Kandakoglu, A., Deha Er, I. (2009) Structuring fuzzy integrated multi-stages
evaluation model on academic personnel recruitment in MET institutions. Expert Systems
with Applications, 36(3), 6918-6927.
Chang, C. L., Hsu, C. H. (2009) Multi-criteria analysis via the VIKOR method for
prioritizing land-use restraint strategies in the Tseng-Wen reservoir watershed. Journal of
Environmental Management, 90(11), 3226-3230.
Chen, C. T. (2000) Extensions of the TOPSIS for group decision making under fuzzy
environment. Fuzzy Sets and Systems, 114(1), 1-9.
Chen, L. Y., Wang, T. C. (2009) Optimizing partners’ choice in IS/IT outsourcing projects:
The strategic decision of fuzzy VIKOR. International Journal of Production Economics,
120(1), 233-242.
Chien, C. F., Chen, L. F. (2008) Data mining to improve personnel selection and enhance
human capital: A case study in high technology industry. Expert Systems with
Applications, 34(1), 280-290.

500
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

Chien, C. F., Chen, L. F. (2007) Using rough set theory to recruit and retain high-potential
talents for semiconductor. IEEE Transactions on Semiconductor Manufacturing, 20(4),
528-541.
Earl, M. (2001) Knowledge Management Strategies: Toward a taxonomy. Journal of
Management Information System, 18(1), 215-233.
Ertugrul Karsak, E. (2000) A fuzzy multiple objective programming approach for personnel
selection. IEEE International Conference on Systems, Man, and Cybernetics, 3, 2007-
2012.
Fan, Z. P., Feng, B., Jiang, Z. Z., Fu, N. (2009) A method for member selection of R&D
teams using the individual and collaborative information. Expert Systems with
Applications, 36(4), 8313-8323.
Feng, B., Jiang, Z. Z., Fan, Z. P., Fu, N. (2010) A method for member selection of cross-
functional teams using the individual and collaborative performances. European Journal
of Operational Research, 203(3), 652-661.
Goumas, M., Lygerou, V. (2000) An extension of the PROMETHEE method for decision
making in fuzzy environment: Ranking of alternative energy exploitation projects.
European Journal of Operational Research, 123(3), 606-613.
Gungor, Z., Serhadloglu, G., Kesen, S. E. (2009) A fuzzy AHP approach to personnel
selection problem. Applied Soft Computing, 9(2), 641-646.
Herrera-Viedma, E., Herrera, F., Martínez, L., Herrera, J. C., López, A. G. (2004)
Incorporating filtering techniques in a fuzzy linguistic multi-agent model for information
gathering on the web. Fuzzy Sets and Systems, 148(1), 61-83.
Herrera-Viedma, E., Cordon, O., Luque, M. A. G., M. Munoz, A. (2003) A model of fuzzy
linguistic IRS based on multigranular linguistic information. International Journal of
Approximate Reasoning, 34(2-3), 221-239.
Herrera, F., Herrera-Viedma, E. (2000) Linguistic decision analysis: Steps for solving
decision problems under linguistic information. Fuzzy Sets and Systems, 115(1), 67-82.
Herrera, F., Martinez, L. (2000) A 2-tuple fuzzy linguistic representation model for
computing with words. IEEE Transactions on Fuzzy Systems, 8(6), 746–752.
Herrera, F., Martinez, L. (2001) A model based on linguistic 2-tuples for dealing with
multigranular hierarchical linguistic contexts in multi-expert decision-making. IEEE
Transactions on Systems, Man, and Cybernetics Part B: Cybernetics, 31(2), 227-234.
Herrera, F., Martinez, L., Sanchez, P. J. (2005) Managing non-homogeneous information in
group decision making. European Journal of Operational Research, 166(1), 115-132.
Korvin, A. D., Shipley, M. F., Kleyle, R. (2002) Utilizing fuzzy compatibility of skill sets for
team selection in multi-phase projects. Journal of Engineering and Technology
Management, 19(3-4), 307-319.
Lai, J. Y., Wang, C. T., Chou, C. Y. (2008) How Knowledge Map and Personalization Affect
Effectiveness of KMS in High-Tech Firms. Proceedings of the 41st Hawaii International
Conference on System Sciences, 355-355.
Opricovic, S. (1998) Multicriteria Optimization of Civil Engineering Systems. Faculty of
Civil Engineering, Belgrade.
Opricovic, S., Tzeng, G. H. (2003) Fuzzy multicriteria model for post-earthquake land-use
planning. Natural Hazards Review, 4(2), 59-64.
Opricovic, S., Tzeng, G. H. (2002) Multicriteria planning of post-earthquake sustainable
reconstruction. Computer-Aided Civil and Infrastructure Engineering, 17(3), 211-220.
Polanyi, M. (1996) The Tacit Dimension. Doubleday, Garden City.
Pyo, S. (2005) Knowledge map for tourist destinations—needs and implications. Tourism
Management, 26(4), 583-594.

501
 C.-T. Chen et al. / Asia Pacific Management Review 16(4) (2011) 491-502

Taft, D. (2000) Stopping Knowledge Overflow—Knowledge Management Tools Still in

Developmental-Phase. Computer Reseller News, 14-14.
Vail, E. F. (1999) Mapping organizational knowledge. Knowledge management review, 8,
10-15.
Rouse, W.B., Thomas, B.S., Boff, K.R. (1998) Knowledge maps for knowledge mining:
Application to R&D/technology management, IEEE Transactions on Systems, Man, and
Cybernetics, Part C: Applications and Reviews, 28(3), 309-317.
Shaw, R. S. (2010) A study of learning performance of e-learning materials design with
knowledge maps. Computers & Education, 54(1), 253-264.
Stain, E. W., Zwass, V. (1995) Actualizing Organizational Memory with information system.
Information System Research, 16(2), 85-117.
Tai, W. S., Chen, C. T. (2009) A new evaluation model for intellectual capital based on
computing with linguistic variable. Expert Systems with Applications, 36(2), 3483-3488.
Wei, Y. H. (2007) A Study on Fuzzy Evaluation Model Based on Non-Monetary Value of
Human Resources. International Conference on Wireless Communications, Networking
and Mobile Computing, 4515-4520.
Woo, J. H., Clayton, M. J., Johnson, R. E., Flores, B. E., Ellis, C. (2004) Dynamic
Knowledge Map: Reusing experts’ tacit knowledge in the AEC industry. Automation in
Construction, 13(2), 203-207.
Xu, Z. S. (2005) Deviation measures of linguistic preference relations in group decision
making. Omega, 33(3), 249-254.

502
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.