Applied Soft Computing 9 (2009) 457–461

Contents lists available at ScienceDirect

Applied Soft Computing

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

Extension of fuzzy TOPSIS method based on interval-valued fuzzy sets

Behzad Ashtiani a,*, Farzad Haghighirad b, Ahmad Makui a, Golam ali Montazer b
a
Department of Industrial Engineering, Iran University of Science and Technology, Narmak, 16846-13114, Tehran, Iran
b
Department of Industrial Engineering, Tarbiat Modares University, Iran

A R T I C L E I N F O A B S T R A C T

Article history: Decision making is one of the most complex administrative processes in management. In circumstances
Received 20 February 2007 where the members of the decision making team are uncertain in determining and defining the decision
Received in revised form 20 May 2008 making criteria, fuzzy theory provides a proper tool to encounter with such uncertainties. However, if
Accepted 26 May 2008
decision makers cannot reach an agreement on the method of defining linguistic variables based on the
Available online 17 July 2008
fuzzy sets, the interval-valued fuzzy set theory can provide a more accurate modeling. In this paper the
interval-valued fuzzy TOPSIS method is presented aiming at solving MCDM problems in which the
Keywords:
weights of criteria are unequal, using interval-valued fuzzy sets concepts.
TOPSIS
Multi criteria decision making
ß 2008 Elsevier B.V. All rights reserved.
Interval-valued fuzzy set

1. Introduction arithmetic operations, which leads to a fuzzy relative closeness for

each alternative.
TOPSIS (Technique for Order Preference by Similarity to Ideal Except for Wang and Elhag [10], our survey shows that the vital
Solution) is a popular approach to multiple criteria decision shortcoming in the other mentioned methods are the loss of
making (MCDM) problems that was proposed by Hwang and Yoon information (defuzzification) in initial steps of their procedure.
[1,2]. This method has been widely used in the literature (Abo- Wang and Elhag’s fuzzy TOPSIS method is based on alpha level sets
sinna and Amer [3], Agrawal et al. [4], Chen and Tzeng [5]). TOPSIS and the fuzzy extension principle, which compute the fuzzy
is a multiple criteria method to identify solution from a set of finite relative closeness of each alternative by solving the Non-linear
alternatives. The basic principle is that the chosen alternative programming models. Final ranking is obtained by defuzzifying the
should have the shortest distance from the positive ideal solution fuzzy relative closeness values.
and the farthest distance from the negative ideal solution. In the In this paper we develop an interval-valued fuzzy TOPSIS
TOPSIS, the performance ratings and the weights of the criteria are (IVF-TOPSIS) to solve MCDM problems in which the performance
given as crisp values. rating values as well as the weights of criteria are linguistics terms
Among many cases, crisp data are inadequate to model real life which can be expressed in interval-valued fuzzy (IVFN) numbers.
situations. Chen [6] extends the TOPSIS method to fuzzy group The remaining of this paper is organized as follows. In the next
decision making situations by considering triangular fuzzy section, we will briefly introduce the TOPSIS method. Section 3
numbers and defining crisp Euclidean distance between two fuzzy illustrates interval-valued fuzzy sets (IVFS). Section 4 describes
numbers. Tsaur et al. [7] convert fuzzy MCDM problem into a crisp developed TOPSIS method to solve interval-valued fuzzy MCDM
one via centroid defuzzification and then solve the non-fuzzy problems. Section 5 investigates a numerical example including an
MCDM problem using the TOPSIS method. Chu and Lin [8] changed application to select a manager for R&D department in a company.
fuzzy MCDM problem into a crisp one. Differing from the others, The paper is concluded in Section 6.
they first derive the membership functions of all the weighted
ratings in a weighted normalized decision matrix and then by 2. TOPSIS method
defuzzifying, convert them to crisp values. Triantaphyllou and Lin
[9] develop a fuzzy version of TOPSIS method based on fuzzy Yoon and Hwang [1,2], introduced the TOPSIS method based on
the idea that the best alternative should have the shortest distance
from an ideal solution. They assumed that if each attributes takes
monotonically increasing or decreasing variation, then it is easy to
* Corresponding author. Tel.: +9821 73913000.
E-mail addresses: [email protected], [email protected]
define an ideal solution. Such a solution is composed of all the best
(B. Ashtiani), [email protected] (F. Haghighirad), [email protected] attributes values achievable, while the worst solution is composed
(A. Makui), [email protected] (G.a. Montazer). of all worst attribute values achievable.

1568-4946/$ – see front matter ß 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.asoc.2008.05.005
458 B. Ashtiani et al. / Applied Soft Computing 9 (2009) 457–461

Suppose a multi criteria decision making problem having n valued fuzzy numbers (IVFN) and gave their extended operations.
alternatives, A1, A2, . . ., An and m criteria, C1, C2, . . ., Cm. Each Interval-valued fuzzy sets have been widely used in real-world
alternative is evaluated with respect to the m criteria. All the applications. For instance, Kohout and Bandler [19] in a CLINAID
values/ratings are assigned to alternatives with respect to decision system, Sambuc [12] in thyrodian pathology, Gorzlczany [14] and
matrix denoted by X(xij)nm. Let W ¼ ðw1 ; w2 ; . . . ; wm Þ be the Bustine [20] in approximate reasoning, Turksen [21,22] in interval-
P
weight vector of criteria, satisfying m j¼1 w j ¼ 1. valued logic and in preference modeling [15]. Based on definition of
The TOPSIS method consists of the following steps: interval-valued fuzzy set in [14], an interval-valued fuzzy set A
defined on (1,+1) is given by:
i. Normalize the decision matrix: the normalization of the A ¼ fðx; ½mLA ðxÞ; mU
A ðxÞg
decision matrix is done using the following transformation
mLA ; mUA : X ! ½0; 1 8 x 2 X; mLA  mUA
for each rij. (8)
m̄A ðxÞ ¼ ½mLA ðxÞ; mUA ðxÞ
xi j
ri j ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn ; i ¼ 1;    ; n; j ¼ 1;    ; m: (1) A ¼ fðx; m̄A ðxÞÞg; x 2 ð1; 1Þ
2
k¼1 xk j
where mLA ðxÞ is the lower limit of degree of membership and mU A ðxÞ
Multiply the columns of the normalized decision matrix by is the upper limit of degree of membership.
the associated weights. The weighted and normalized decision Fig. 1 illustrates the membership value at x0 of interval-valued
matrix is obtained as: fuzzy set A. Thereby, the minimum and maximum membership
Vi j ¼ w j  ri j ; i ¼ 1;    ; n j ¼ 1;    ; m (2) value of x0 are mLA ðxÞ and mU A ðxÞ, respectively.
Given two interval-valued fuzzy numbers Nx ¼ ½Nx ; Nxþ  and
where w j represents the weight of the jth criterion. M y ¼ ½My ; Myþ , according to [23,24], we have:
ii. Determine the ideal and negative ideal alternatives: the ideal
Definition 1. If  2 (+,,,), then N  Mðx  yÞ ¼ ½Nx  My ; Nxþ  Myþ 
and negative ideal alternatives are determined, respectively, as
follows: Definition 2. The Normalized Euclidean distance between Ñ and M̃
A ¼ fv1 ; v2 ;    ; vm g is as follows:
    vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 3
u1 X
DðÑ; M̃Þ ¼ t
max vi j j j 2 Vb ; min vi j j j 2 Vc (3) 2
¼ ½ðNxi  Myi Þ2 þ ðNxþi  Myþi Þ 
j j 6 i¼1
A ¼ fv  
1 ; v2 ;    ; vm g
   
¼ min vi j j j 2 Vb ; max vi j j j 2 Vc (4)
j j
4. The proposed interval-valued fuzzy TOPSIS
where Vb is the set of benefit criteria and Vc is the set of cost
criteria. In fuzzy MCDM problems, performance rating values and
iii. Obtain the distance of the existing alternatives from ideal and relative weights are usually characterized by fuzzy numbers. A
negative ideal alternatives: the two Euclidean distances for fuzzy number is a convex fuzzy set, defined by a given interval of
each alternative are, respectively, calculated as: real numbers, each with a membership value between 0 and 1.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Considering the fact that, in some cases, determining precisely of
uX
u m this value is difficult, the membership value can be expressed as an
Si ¼ t ðvi j  vj Þ2 ; i ¼ 1;    ; n
þ
(5)
j¼1 interval, consisting real numbers. In this paper criteria values’ as
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX well as criteria weights’, are considered as linguistic variables. The
u m
S
i ¼
t ðvi j  v Þ2 ; i ¼ 1;    ; n: (6) concept of linguistic variable is very useful in dealing with
j
j¼1 situations that are too complex or ill-defined to be reasonably
described in conventional quantitative expressions [11]. These
iv. Calculate the relative closeness to the ideal alternatives: the
linguistic variables can be converted to triangular interval-valued
relative closeness to the ideal alternatives can be defined as:
fuzzy numbers as depicted in Tables 1 and 2.
S
i Let X̃ ¼ ½x̃i j nm be a fuzzy decision matrix for a multi criteria
RCi ¼ ; i ¼ 1; 2;    ; m; 0  RCi  1: (7)
Sþ
i þ S
i decision making problem in which A1, A2, . . ., An are n possible
alternatives and C1, C2, . . ., Cm are m criteria. So the performance of
v. Rank the alternatives: according to the relative closeness to the
alternative Ai with respect to criterion Cj is denoted as x̃i j . As
ideal alternatives, the bigger is the RCi, the better is the
illustrated in Fig. 2, x̃i j and w̃ j are expressed in triangular interval-
alternative Ai.
valued fuzzy numbers.

ðx1 ; x2 ; x3 Þ
3. Interval-valued fuzzy sets x̃ ¼
ðx01 ; x2 ; x03 Þ
In fuzzy sets theory, it is often difficult for an expert to exactly
quantify his or her opinion as a number in interval [0,1]. Therefore, it
is more suitable to represent this degree of certainty by an interval.
Sambuc [12] and Grattan [13] noted that the presentation of a
linguistic expression in the form of fuzzy sets is not enough. Interval-
valued fuzzy sets were suggested for the first time by Gorzlczany
[14] and Turksen [15]. Also Cornelis et al. [16] and Karnik and
Mendel [17] noted that the main reason for proposing this new
concept is the fact that in the linguistic modeling of a phenomenon,
the presentation of the linguistic expression in the form of ordinary
fuzzy sets is not clear enough. Wnag and Li [18] defined interval- Fig. 1. Interval-valued fuzzy set.
 B. Ashtiani et al. / Applied Soft Computing 9 (2009) 457–461 459

Table 1
The normalization logic used in (10) is the same with which
Definitions of linguistic variables for the ratings
is used in deterministic problems.
Very Poor (VP) [(0,0);0;(1,1.5)]
Poor (P) [(0,0.5);1;(2.5,3.5)]
Hence, the normalized matrix R̃ ¼ ½r̃i j nm can be obtained.
Moderately Poor (MP) [(0,1.5);3;(4.5,5.5)] ii. By considering the different importance of each criterion, we
Fair (F) [(2.5,3.5),5,(6.5,7.5)] can construct the weighted normalized fuzzy decision matrix
Moderately Good (MG) [(4.5,5.5),7,(8,9.5)] as: Ṽ ¼ ½ṽi j nm where ṽi j ¼ r̃i j  w̃ j . According to Defintion 1,
Good (G) [(5.5,7.5),9,(9.5,10)]
the multiply operator can be applied as:
Very Good (VG) [(8.5,9.5),10,(10,10)]
ṽi j ¼ ½ðr̃1i j  w̃1 j ; r̃01i j  w̃01 j Þ; r̃2i j  w̃2 j ; ðr̃03i j  w̃03 j ; r̃3i j  w̃3 j Þ
¼ ½ðgi j ; g0i j Þ; hi j ; ðli j ; l0i j Þ (11)

Table 2 iii. Ideal and negative ideal solution can be defined as:
Definitions of linguistic variables for the importance of each criterion
Aþ ¼ ½ð1; 1Þ; 1; ð1; 1Þ; j 2 Vb
Very low (VL) [(0,0);0;(0.1,0.15)] (12)
Low (L) [(0,0.05);0.1;(0.25,0.35)]
A ¼ ½ð0; 0Þ; 0; ð0; 0Þ; j 2 Vc
Medium low (ML) [(0,0.15);0.3;(0.45,0.55)]
iv. Normalized Euclidean distance can be calculated using Defini-
Medium (M) [(0.25,0.35),0.5,(0.65,0.75)]
Medium high (MH) [(0.45,0.55),0.7,(0.8,0.95)] tion 2 as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
High (H) [(0.55,0.75),0.9,(0.95,1)] u 3
Very high (VH) [(0.85,0.95),1,(1,1)] u1 X
D ðÑ; M̃Þ ¼ t

½ðN xi  Myi Þ 
 2
3 i¼1
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 3
u1 X
The x̃ can be also demonstrated as x̃ ¼ ½ðx1 ; x01 Þ; x2 ; ðx03 ; x3 Þ. It is D ðÑ; M̃Þ ¼ t
þ
½ðNþ xi  Myi Þ 
þ 2

worth noting, the use of interval value numbers gives an 3 i¼1

opportunity to experts to define lower and upper bounds values
as an interval for matrix’s elements and weights of criteria. Also in where D ðÑ; M̃Þ and Dþ ðÑ; M̃Þ are the primary and secondary
a group decision environment with K persons, the importance of distant measure, respectively. Thereby, distance of each
the criteria and the rating of alternatives with respect to each alternative from the ideal alternative ½Dþ i1
; Dþ i2
 can be currently
criterion can be calculated as: calculated, where:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 Xm
1
x̃i j ¼ ½x̃1i j þ x̃2i j þ    þ x̃kij  Dþ ¼ ½ðg  1Þ2 þ ðhi j  1Þ2 þ ðli j  1Þ2 
K (9)
i1
j¼1
3 ij
1 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (13)
w̃i j ¼ ½w̃1i j þ w̃2i j þ    þ w̃kij  Xm
1 0
K Dþ ¼ ½ðg  1Þ
2
þ ðh  1Þ 2
þ ðl 0
 1Þ
2

i j ij
i2
j¼1
3 ij
Eq. (9) represents the average values of xij and wi j denoted by
experts, where (+) is the sum operator and is applied to the
Similarly, the separation from the negative ideal solution is
interval-valued fuzzy numbers as defined in Definition 1. So the
given by ½D 
i2 ; Di1 , where:
output is also an interval value fuzzy number. r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Now the proposed approach to develop the TOPSIS for interval- Xm
1
Di1 ¼ ½ðg  0Þ2 þ ðhi j  0Þ2 þ ðli j  0Þ2 
valued fuzzy data (IVF-TOPSIS) can be defined as follows: j¼1
3 ij
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (14)
Xm
1 0 2 2
i. Given x̃i j ¼ ½ðai j ; a0i j Þ; bi j ; ðc0i j ; ci j Þ, the normalized performance Di2 ¼ ½ðgi j  0Þ þ ðhi j  0Þ2 þ ðl0i j  0Þ 
rating as an extension to Chen [6] can be calculated as: j¼1
3
" 0
! 0
!#
ai j ai j bi j ci j ci j Eqs. (13) and (14) are employed to determine the distance
r̃i j ¼ ; ; þ ; þ ; þ ; i ¼ 1; . . . ; n; j 2 Vb
cþj cþj cj cj cj from ideal and negative ideal alternatives in interval values. In
"  !  !#
aj aj aj aj aj this way we lose less information (data values) than just
r̃i j ¼ ; ; ; ; ; i ¼ 1; . . . ; n; j 2 Vc (10) converting immediately to crisp values.
a0 a ij ij b ij c c0
ij ij
v. The relative closeness can be calculated as follows:
cþj ¼ Max ci j ; j 2 Vb
i Di2 Di1
RC1 ¼  ; RC2 ¼ (15)
aj ¼ Min a0i j ; j 2 Vc Dþ
i2 þ Di2 Dþ 
i1 þ Di1
i

The final values of RCi are determined as:

RC1 þ RC2
RCi ¼ (16)
2

As a summary, the Interval-valued fuzzy TOPSIS can be summed

up as follows:

Normalize fuzzy decision matrix X̃ ¼ ½x̃i j nm by Eqs. (10) and

(11).
Determine the ideal solution and the negative ideal solution by
Eqs. (12).
Calculate normalize Euclidean distances by Eqs. (13) and (14).
Compute the fuzzy relative closeness of each alternative by using
Fig. 2. Interval-valued triangular fuzzy number. each pair of separations (15).
460 B. Ashtiani et al. / Applied Soft Computing 9 (2009) 457–461

Table 3 Determine the relative closeness by Eq. (16).

The importance of criterion
Rank alternatives in terms of their relative closeness’s.
DM1 DM2 DM3

C1 VH VH VH 5. The application of the extended method in solving problems

C2 H H MH
C3 H MH MH
C4 VH H VH Suppose that a Telecommunication Company intends to choose
C5 M MH M a manager for R&D department from four volunteers named A1, A2,
A3, A4. The decision making committee assesses the four concerned
volunteers based on five criteria which follow:

Table 4 a) proficiency in identifying research areas (C1),

Decision makers’ assessments based on each criterion b) proficiency in administration (C2),
c) personality (C3),
Decision makers Volunteer Criterion
d) past experience (C4) and
DM3 DM2 DM1 e) self-confidence (C5)
C1 A1 MG G VG
A2 MG G G
A3 VG G VG The number of the committee members is three, labeled as DM1,
A4 MG VG G DM2, DM3 respectively.
Each decision maker has presented his assessment based on
C2 A1 F MG VG
A2 MG VG VG linguistic variable for rating performance and importance of each
A3 VG G MG criterion by a linguistic variable as depicted in Tables 3 and 4,
A4 VG F F respectively.
C3 A1 G G VG We will proceed to solve the problem using the interval-valued
A2 G VG VG fuzzy TOPSIS. Table 5 shows the final judgment of the decision
A3 VG MG G makers through applying Eq. (9).
A4 MG MG F
Afterwards the decision matrix is normalized by using Eq. (10).
C4 A1 VG G VG Table 6 depicts the normalized decision matrix.
A2 MG VG VG Table 7 shows the weighted normalized matrix. Euclidean
A3 VG VG G
distance from the ideal and negative ideal alternatives are
A4 VG F G
calculated using (13) and (14) formulas, respectively. The results
C5 A1 VG VG VG have been depicted in Table 8. As demonstrated, the distance from
A2 G MG MG
ideal and negative ideal alternatives are determined as an interval.
A3 MG G G
A4 F G MG Applying Eq. (15), the interval relative closeness obtained and
the results are depicted in Table 9.

Table 5
The interval-valued fuzzy decision matrix and weights

C1 C2 C3 C4 C5

A1 [(6.17,7.5),8.67,(9.17,9.83)] [(5.17,6.17);7.33;(8.17,9)] [(6.5,8.17);9;(9.33,9.83)] [(7.5,8.83);9.67;(9.83,10)] [(8.5,9.5);10;(10,10)]

A2 [(5.17,6.83),8.33,(9,9.83)] [(7.17,8.17);9;(9.33,9.83)] [(7.5,8.83);9.67;(9.83,10)] [(7.17,8.17);9;(9.33,9.83)] [(4.83,6.17);7.67;(8.5,9.67)]
A3 [(7.5,8.83);9.67;(9.83,10)] [(6.17,7.5);8.67;(9.17,9.83)] [(6.17,7.5);8.67;(9.17,9.83)] [(7.5,8.83);9.67;(9.83,10)] [(5.17,6.83);8.33;(9,9.83)]
A4 [(6.17,7.5);8.67;(9.17,9.83)] [(4.5,5.5);6.67;(7.67,8.33)] [(3.83,4.83);6.33;(7.5,8.83)] [(5.5,6.83);8;(8.67,9.17)] [(4.17,5.5);7;(8,9)]
Weight [(0.85,0.95);1;(1,1)] [(0.52,0.68);0.83;(0.9,0.98)] [(0.48,0.62);0.77;(0.85,0.97)] [(0.75,0.88);0.97;(0.98,1)] [(0.32,0.42);0.57;(0.7,0.82)]

Table 6
Normalized decision matrix

C1 C2 C3 C4 C5

A1 [(0.62,0.75);0.87;(0.92,0.98)] [(0.52,0.62);0.73;(0.82,0.9)] [(0.65,0.82);0.93;(0.97,1)] [(0.75,0.88);0.97;(0.98,1)] [(0.85,0.95);1;(1,1)]

A2 [(0.52,0.68);0.83;(0.9,0.98)] [(0.72,0.82);0.9;(0.93,0.98)] [(0.75,0.88);0.97;(0.98,1)] [(0.72,0.82);0.9;(0.93,0.98)] [(0.48,0.62);0.77;(0.85,0.97)]
A3 [(0.75,0.88);0.97;(0.98,1)] [(0.62,0.75);0.87;(0.92,0.98)] [(0.62,0.75);0.87;(0.92,0.98)] [(0.75,0.88);0.97;(0.98,1)] [(0.52,0.68);0.83;(0.9,0.98)]
A4 [(0.62,0.75);0.87;(0.92,0.98)] [(0.45,0.55);0.67;(0.77,0.83)] [(0.38,0.48);0.63;(0.75,0.88)] [(0.55,0.68);0.8;(0.87,0.92)] [(0.42,0.55);0.7;(0.8,0.9)]

Table 7
Weighted normalized decision matrix

C1 C2 C3 C4 C5

A1 [(0.52,0.71);0.87;(0.92,0.98)] [(0.27,0.42);0.61;(0.74,0.89)] [(0.31,0.5);0.72;(0.82,0.97)] [(0.56,0.78);0.93;(0.97,1)] [(0.27,0.4);0.57;(0.7,0.82)]

A2 [(0.44,0.65);0.83;(0.9,0.98)] [(0.37,0.56);0.75;(0.84,0.97)] [(0.36,0.54);0.74;(0.84,0.97)] [(0.54,0.72);0.87;(0.92,0.98)] [(0.15,0.26);0.43;(0.6,0.79)]
A3 [(0.64,0.84);0.97;(0.98,1)] [(0.32,0.51);0.72;(0.83,0.97)] [(0.3,0.46);0.66;(0.78,0.95)] [(0.56,0.78);0.93;(0.97,1)] [(0.16,0.28);0.47;(0.63,0.8)]
A4 [(0.52,0.71);0.87;(0.92,0.98)] [(0.23,0.38);0.56;(0.69,0.82)] [(0.19,0.3);0.49;(0.64,0.85)] [(0.41,0.6);0.77;(0.85,0.92)] [(0.13,0.23);0.4;(0.56,0.74)]
 B. Ashtiani et al. / Applied Soft Computing 9 (2009) 457–461 461

Table 8 [2] K. Yoon, System selection by multiple attribute decision making, Ph.D. disserta-
The distance from the ideal solution and negative-ideal solution tions, Kansas State University, Manhattan, Kansas, 1980.
[3] M.A. Abo-Sinna, A.H. Amer, Extensions of TOPSIS for multi objective large-scale
½Dþ
i1
; Dþ
i2
 ½D 
i2 ; Di1  nonlinear programming problems, Applied Mathematics and Computation 162
(2005) 243–256.
A1 [1.56,1.96] [3.6,3.62] [4] V.P. Agrawal, V. Kohli, S. Gupta, Computer aided robot selection: the multiple
A2 [1.63,2.01] [3.54,3.6] attribute decision making approach, International Journal of Production Research
A3 [1.51,1.92] [3.66,3.7] 29 (1991) 1629–1644.
A4 [2.11,2.37] [3.05,3.2] [5] M.F. Chen, G.H. Tzeng, Combining grey relation and TOPSIS concepts for selecting
an expatriate host country, Mathematical and Computer Modeling 40 (2004)
1473–1490.
[6] C.T. Chen, Extensions of the TOPSIS for group decision-making under fuzzy
Table 9 environment, Fuzzy Sets and Systems 114 (2000) 1–9.
The interval of relative closeness [7] S.H. Tsaur, T.Y. Chang, C.H. Yen, The evaluation of airline service quality by fuzzy
MCDM, Tourism Management 23 (2002) 107–115.
A1 [0.65,0.7]
[8] T.C. Chu, Y.C. Lin, A fuzzy TOPSIS method for robot selection, The Inter-
A2 [0.64,0.68]
national Journal of Advanced Manufacturing Technology 21 (2003) 284–
A3 [0.66,0.71] 290.
A4 [0.57,0.59] [9] E. Triantphyllou, C.T. Lin, Development and evaluation of five fuzzy multi attribute
decision making methods, International Journal of Approximate Reasoning 14
(1996) 281–310.
Finally, with using Eq. (16), the value of each alternative for final [10] Y.M. Wang, T.M.S. Elhag, Fuzzy TOPSIS method based on alpha level sets with an
application to bridge risk assessment, Expert Systems with Applications 31
ranking will be: (2005) 309–319.
RC1 ¼ 0:673 [11] L. Zadeh, The concept of a linguistic variable and its application to approximate
RC2 ¼ 0:664 reasoning - I, Information Science 8 (1975) 199–249.
[12] R. Sambuc, Fonctions f-floues. Application à l’aide au diagnostic en pathologie
RC3 ¼ 0:683 thyroidienne, Ph.D. Thesis, University of Marseille 1975.
RC4 ¼ 0:583 [13] I. Grattan-Guinness, Fuzzy membership mapped onto interval and many-valued
quantities, Z. Math. Logik Grundlag. Mathe. 22 (1975) 149–160.
Therefore, the final ranking is: [14] M.B. Gorzalczany, A method of inference in approximate reasoning based on
A3 > A1 > A2 > A4 interval-valued fuzzy sets, Fuzzy Sets and Systems 21 (1987) 1–17.
[15] I.B. Turksen, Interval-valued strict preference with Zadeh triples, Fuzzy Sets and
Systems 78 (1996) 183–195.
[16] C. Cornelis, G. Deschrijver, E.E. Kerre, Advances and challenges in interval-valued
6. Conclusion fuzzy logic, Fuzzy Sets and Systems 157 (2006) 622–627.
[17] N.N. Karnik, J.M. Mendel, Operations on type-2 fuzzy sets, Fuzzy Sets and Systems
It is argued that if a fuzzy MCDM problem is defuzzified into a 122 (2001) 327–348.
[18] G. Wang, X. Li, The applications of interval-valued fuzzy numbers and
crisp one in initial steps, then the advantage of collecting fuzzy data interval-distribution numbers, Fuzzy Sets and Systems 98 (1998) 331–
becomes unapparent. Based on this fact, we have developed a fuzzy 335.
TOPSIS method for dealing with problems, in which criteria values’ [19] L.J. Kohout, W. Bandler, Fuzzy interval inference utilizing the checklist paradigm
and BK-relational products, in: R.B. Kearfort, et al. (Eds.), Applications of Interval
are interval-valued fuzzy numbers. The proposed fuzzy TOPSIS
Computations, Kluwer, Dordrecht, 1996, pp. 291–335.
method combines the TOPSIS method for crisp MCDM with the fuzzy [20] H. Bustince, Conjuntos Intuicionistas e Intervalo valorados Difusos: Propiedades y
extension principles and performs defuzzification in the final step of Construcci_on, Relaciones Intuicionistas Fuzzy, Thesis, Universidad P_ublica de
Navarra, 1994.
decision analysis process. The other aggregation functions can be
[21] I.B. Turksen, Interval-valued fuzzy sets based on normal forms, Fuzzy Sets and
used to pool the fuzzy ratings of decision-makers in the proposed Systems 20 (1986) 191–210.
method. Utilizing the proposed IVF-TOPSIS method, a manager [22] I.B. Turksen, Interval-valued fuzzy sets and compensatory, Fuzzy Sets and Sys-
selection problem was examined and the results are demonstrated. tems 51 (1992) 295–307.
[23] D.H. Hong, S. Lee, Some algebraic properties and a distance measure for interval-
valued fuzzy numbers, Information Sciences 148 (2002) 1–10.
References [24] P. Grzegorzewski, Distances between intuitionistic fuzzy sets and/or interval-
valued fuzzy sets based on the Hausdorff metric, Fuzzy Set and Systems 148
[1] C.L. Hwang, K. Yoon, Multiple Attributes Decision Making Methods and (2004) 319–328.
Applications, Springer, Berlin Heidelberg, 1981.