Anda di halaman 1dari 11


Multi-criteria decision-making method based on a cross-entropy with interval

neutrosophic sets
Zhang-peng Tian , Hong-yu Zhang , Jing Wang , Jian-qiang Wang and Xiao-hong Chen
School of Business, Central South University, Changsha, PR China


In this paper, two optimisation models are established to determine the criterion weights in multi-criteria Received May
decision-making situations where knowledge regarding the weight information is incomplete and the cri- Accepted September
terion values are interval neutrosophic numbers. The proposed approach combines interval neutrosophic
sets and TOPSIS, and the closeness coefficients are expressed as interval numbers. Furthermore, the relative Multi-criteria
likelihood-based comparison relations are constructed to determine the ranking of alternatives. A fuzzy decision-making; interval
cross-entropy approach is proposed to calculate the discrimination measure between alternatives and the neutrosophic sets;
absolute ideal solutions, after a transformation operator has been developed to convert interval neutro- cross-entropy; mathematical
sophic numbers into simplified neutrosophic numbers. Finally, an illustrative example is provided, and a programming; TOPSIS
comparative analysis is conducted between the approach developed in this paper and other existing meth-
ods, to verify the feasibility and effectiveness of the proposed approach.

1. Introduction falsity membership simultaneously, it is more practical and flex-

Fuzzy sets (FSs), which were proposed by Zadeh (1965), are ible than FSs and IFSs in dealing with uncertain, incomplete
regarded as a comprehensive tool for solving multi-criteria and inconsistent information. For the aforementioned exam-
decision-making (MCDM) problems (Bellman & Zadeh, 1970). ple, the reviewers opinion can be presented as x(0.7, 0.2, 0.6)
In order to resolve the uncertainty of non-membership degrees, by means of NSs. However, without a specific description, it
Atanassov (1986) introduced intuitionistic fuzzy sets (IFSs), is hard to apply NSs in actual scientific and engineering situ-
which are an extension of Zadehs FSs. IFSs have been widely ations. Hence, single-valued neutrosophic sets (SVNSs), which
applied in solving MCDM problems to date (Chen & Chang, are an extension of NSs, were introduced by Wang et al. (2010).
2015; Chen, Cheng, & Chiou, 2016; Wang et al., 2014; Yue, 2014). Subsequently, the similarity and entropy measures (Majum-
Moreover, interval-valued intuitionistic fuzzy sets (IVIFSs) dar & Samant, 2014), the correlation coefficient (Ye, 2013)
(Atanassov & Gargov, 1989) were proposed, which are an exten- and the cross-entropy (Ye, 2014c) of SVNSs have been devel-
sion of FSs and IFSs. In recent years, MCDM problems with oped. Additionally, Ye (2014a) introduced simplified neutro-
IVIFSs have attracted much attention from researchers (Chen, sophic sets (SNSs), and Peng, Wang, Wang, Zhang, and Chen
2014; Liu, Shen, Zhang, Chen, & Wang, 2015; Tan et al., 2014; (2015) and Peng, Wang, Zhang, and Chen (2014) defined their
Wan & Dong, 2014). Furthermore, the TOPSIS method, pro- novel operations and aggregation operators. Finally, further
posed by Hwang and Yoon (1981), has also been used for solving extensions of NSs, such as interval neutrosophic sets (INSs)
MCDM problems (Cao, Wu, & Liang, 2015; Yue, 2014; Zhang (Wang et al., 2005), bipolar neutrosophic sets (BNSs) (Deli,
& Yu, 2012). Moreover, fuzzy linear programming models have Ali, & Smarandache, 2015) and multivalued neutrosophic sets
been constructed to address MCDM problems with incomplete (MVNSs) (Peng, Wang, Wu, Wang, & Chen, 2015; Wang & Li,
criterion weight information (Chen, 2014; Dubey, Chandra, & 2015), have also been proposed.
Mehra, 2012; Wan, Wang, Lin & Dong, 2015; Wan, Xu, Wang, & In certain real-life situations, the INS, as a particular exten-
Dong, 2015; Zhang & Yu, 2012). sion of an NS, can be more flexible in assessing objections than
Although the theories of FSs and IFSs have been developed an SNS. Recently, the studies relating to INSs have been focused
and generalised, they cannot deal with all types of uncertain- on particular areas, which can be roughly classified into two
ties in real problems. Indeed certain types of uncertainties, such groups.
as indeterminate and inconsistent information, cannot be man- The first group is based on interval neutrosophic aggregation
aged. For example, FSs and IFSs cannot effectively deal with a sit- operators, such as interval neutrosophic number weighted aver-
uation where a paper is sent to a reviewer, and he or she says it is aging (INNWA) and interval neutrosophic number weighted
70% acceptable and 60% unacceptable, and his or her statement geometric (INNWG) operators (Zhang, Wang, & Chen, 2014),
is 20% uncertain; therefore, some new theories are required. interval neutrosophic number Choquet integral (INNCI) oper-
Since neutrosophic sets (NSs) (Smarandache, 1999) con- ator (Sun et al., 2015), interval neutrosophic number ordered
sider the truth membership, indeterminacy membership and weighted averaging (INNOWA) and interval neutrosophic

CONTACT Jian-qiang Wang

Taylor & Francis
2 Z.-p. TIAN ET AL.

number ordered weighted geometric (INNWG) operators (Ye, can transform each INN into an SNN, a cross-entropy for SNNs
2015) and interval neutrosophic prioritised ordered weighted is proposed. Subsequently, a TOPSIS approach in the context of
aggregation (INPOWA) operator (Liu & Wang, 2015). INSs for solving MCDM problems with incompletely known cri-
The second group is based on interval neutrosophic mea- terion information is developed in Section 4. In Section 5, an
sures (Broumi & Smarandache, 2014a, 2014b; Chi & Liu, 2013; illustrative example is provided and a comparative analysis is
Ye, 2014b; Zhang, Ji, Wang, & Chen 2015a; Zhang et al. 2015b). conducted between the proposed approach and other existing
Specifically, Chi and Liu (2013) extended TOPSIS to an INS methods. Finally, conclusions are drawn in Section 6.
based on a distance measure. Broumi and Smarandache (2014a,
2014b) defined a new cosine similarity measure and a correla- 2. Preliminaries
tion coefficient of an INS. Moreover, Ye (2014b) defined two In this section, some basic concepts and definitions related to
interval neutrosophic similarity measures based on the Ham- INSs, including interval numbers, definitions and operational
ming and Euclidean distances. Zhang et al. (2015b) defined sev- laws of NSs, SNSs and INSs are introduced; these will be utilised
eral interval neutrosophic outranking relations based on ELEC- in the latter analysis.
TRE IV. Finally, Zhang et al. (2015a) proposed an improved
weighted correlation coefficient based on the integrated weight
2.1. Interval numbers
for an INS.
Interval numbers and their operations are of utmost significance
The aforementioned methods are effective when managing
when exploring the operations of INSs. In the following para-
interval neutrosophic MCDM problems; however, they have
graphs, some definitions and operational laws of interval num-
some drawbacks which are outlined below.
bers are provided.
(1) The existing MCDM methods require the criterion weight
information to be completely known (Broumi & Smarandache, Definition 1 (Sengupta & Pal, 2000; Xu, 2008): Let a =
2014b; Ye, 2014b, 2015; Zhang et al., 2014, 2015a, 2015b), or are [aL , aU ] = {x|aL x aU }, and then a is called an interval
supported by fuzzy measures of criteria (Sun et al., 2015). How- number. In particular, a = [aL , aU ] will be degenerated to a real
ever, due to the increasing complexity of MCDM problems, it is number, if aL = aU .
difficult and subjective to provide exact criterion weight infor-
mation or fuzzy measures. (2) The MCDM methods (Broumi & Consider two non-negative interval numbers a = [aL , aU ]
Smarandache, 2014b; Chi & Liu, 2013; Ye, 2014b; Zhang et al., and b = [bL , bU ], where 0 aL x aU and 0 bL x
2015a, 2015b) utilise the weighted measures, such as distance, bU . Subsequently, their operations are defined as follows (Sen-
correlation coefficient and similarity measures, to rank alter- gupta & Pal, 2000; Xu, 2008):
natives; in these processes, the interval neutrosophic informa- (1) a + b = [aL + bL , aU + bU ],
tion is measured with crisp real numbers. However, the means (2) a = [aL , aU ], > 0.
of aggregating interval numbers into real numbers may lead to Definition 2 (Xu & Da, 2003): For any two interval numbers
some operational deficiencies and a large amount of informa- a = [aL , aU ] and b = [bL , bU ], the possibility of a b is for-
tion loss. (3) All of the developed aggregation operators (Liu &
mulated by p(a b) = max{1 max{ b a , 0}, 0}, where

Wang, 2015; Sun et al., 2015; Ye, 2015; Zhang et al., 2014) directly L(a)+L(b)
process the assessment information with INNs. However, these L(a) = aU aL and L(b) = bU bL .
approaches are complex and tedious, and not enough attention
is given to reducing the computational complexity when pro- The possibility degree of a b has the following properties
cessing the evaluation information. (Xu & Da, 2003):
To overcome these disadvantages, in this paper, a novel and (1) 0 p(a b) 1;
comprehensive approach for managing MCDM with INNs is (2) p(a b) = p(b a) = 0.5, if p(a b) = p(b a);
developed. The main novelties and the significant contributions (3) p(a b) + p(b a) = 1;
are summarised below. (4) p(a b) = 0 if aU bL , p(a b) = 1, if aL bU ;
(1) In order to derive the criterion weights, two mathemati- (5) For any interval numbers a, b and c, p(a c) 0.5, if
cal programming models are constructed to determine the opti- p(a b) 0.5 and p(b c) 0.5; p(a c) = 0.5, if
mal weight values, and decision-makers (DMs) may provide and only if p(a b) = p(b c) = 0.5.
incomplete or inconsistent opinions on the weight information.
(2) A novel form of the closeness coefficient with interval val-
ues is derived on the basis of TOPSIS and a cross-entropy of 2.2. NSs and SNSs
INSs. Furthermore, the weighted interval closeness coefficients Due to the influence of subjective factors, it is difficult for DMs
are employed to determine the ranking of alternatives. (3) To to explicitly express preferences if indeterminacy and inconsis-
lower the computational complexity, a modified transforma- tency exist. NS can effectively capture such information (Guo
tion operator is developed for converting INNs into simplified & Sengr, 2014; Mohan, Krishnaveni, & Guo, 2013; Solis &
neutrosophic numbers (SNNs). Nevertheless, the parameters of Panoutsos, 2013).
the transformation operator are not artificially set, but obtained Definition 3 (Smarandache, 1999): Let X be a space of points
through mathematical derivation. (objects) with a generic element in X, denoted by x. An NS A
The rest of the paper is organised as follows. In Section 2, in X is characterised by a truth-membership function tA (x),
interval numbers, as well as the concepts of NSs, SNSs and INSs, an indeterminacy-membership function iA (x) and a falsity-
are briefly reviewed. In Section 3, on the basis of an operator that membership function fA (x). tA (x), iA (x) and fA (x) are real

standard or non-standard subsets of ]0 , 1+ [; that is, tA (x) : Definition 8 Let p, q, r [0, 1] be three fixed numbers; any INN
X ]0 , 1+ [, iA (x) : X ]0 , 1+ [, and fA (x) : X ]0 , can be transformed into an SNN through the operator Hp,q,r :
1+ [. There is no restriction on the sum of tA (x), iA (x) and
fA (x), thus 0 sup tA (x) + sup iA (x) + sup fA (x) 3+ .
Hp,q,r (A) = tAL + pWtA , iLA + qWiA , fAL + rW fA ,
Since it is difficult to apply NSs to practical problems, Ye
(2014a) reduced NSs of non-standard interval numbers into a where WtA = t U tAL , WiA = iU iLA and W fA = f U fAL .
type of SNS of standard interval numbers.
Obviously, Hp,q,r (A) is an SNN, determined with respect to p, q
Definition 4 (Rivieccio, 2008; Ye, 2014a): Let an NS A in X be and r; that is, Hp,q,r (A) is well defined in all value ranges of p, q
characterised by tA (x), iA (x) and fA (x), which are single subin- and r.
tervals/subsets in the real standard [0, 1]; that is, tA (x) : X
[0, 1], iA (x) : X [0, 1], and fA (x) : X [0, 1]. Also, the Example 1: Assume two INNs A = ([0.7, 0.8], [0, 0.1],
sum of tA (x), iA (x) and fA (x) satisfies the condition 0 tA (x) + [0.1, 0.2]) and B = ([0.4, 0.5], [0.2, 0.3], [0.3, 0.4]). Then, the
iA (x) + fA (x) 3. Then, a simplification of A is denoted by following transformation results can be obtained utilising the
A = {(x, tA (x), iA (x), fA (x))|x X}, which is called an SNS operator Hp,q,r :
and is a subclass of NSs. If X = 1, an SNS will be degenerated
to an SNN. Hp,q,r (A) = (0.7 + p 0.1, 0 + q 0.1, 0.1 + r 0.1) and
Definition 5 (Rivieccio, 2008; Ye, 2014a): An SNS A is con- Hp,q,r (B) = (0.4 + p 0.1, 0.2 + q 0.1, 0.3 + r 0.1).
tained in the other SNS B, denoted by A B, if and only if
tA (x) tB (x), iA (x) iB (x) and fA (x) fB (x), for anyx X. It is shown that Hp,q,r (A) and Hp,q,r (B) will be two spe-
The complement set of A, denoted by Ac , is defined as Ac = cific SNNs if the values of p, q and r are given; the method to
{(x, fA (x), iA (x), tA (x))|x X}. determine the parameter values will be discussed in detail in
Section 4.
2.3. INSs
In actual applications, sometimes it is not easy to express 3.2. A cross-entropy measure of SNNs
the truth membership, indeterminacy membership and falsity Shang and Jiang (1997) proposed a fuzzy cross-entropy and a
membership by crisp values, but they may be easily described symmetric discrimination information measure between two
by interval numbers. Wang et al. (2005) further defined INSs. FSs. Vlachos and Sergiadis (2007) then proposed an intuition-
Definition 6 (Rivieccio, 2008; Wang et al., 2005): Let X be a istic fuzzy cross-entropy for IFSs, and Ye (2011) defined a
space of points (objects) with generic elements in X, denoted fuzzy cross-entropy for IVIFSs. Furthermore, the fuzzy cross-
by x. An INS A in X is characterised by a truth-membership entropy has been employed for various purposes including
function tA (x), an indeterminacy-membership function iA (x) deriving criterion weights by constructing mathematical pro-
and a falsity-membership function fA (x). For each point x in gramming models (Zhang & Yu 2012); technical efficiency
X, tA (x) = [tAL , t U ], iA (x) = [iLA , iU ] and fA (x) = [ fAL , f U ], and analysis (Macedo & Scotto, 2014); multi-objective optimisa-
A A A tion (Caballero, Hernndez-Daz, Laguna, & Molina, 2015); and
0 t U + iU + f U 3. In particular, an INS will be reduced to
A A A MCDM problems (Meng & Chen, 2015; Peng, Wang, Wu, et al.,
an SNS, if tAL = t U , iLA = iU and fAL = f U . In addition, if X = 2014; Ye, 2014c; Zhao et al., 2013).
1, an INS will be degenerated to an INN. In a similar manner to the proposals of Ye (2014c) and Vla-
chos and Sergiadis (2007), the following definition of a fuzzy
Definition 7 (Rivieccio, 2008; Wang et al., 2005): An INS A
cross-entropy for SNNs is proposed.
is contained in the other INS B, denoted by A B, if and
only if tAL tBL , t U tBU , iLA iLB , iU iUB , fAL fBL and f U Definition 9 Let A, B SNS, and then the cross-entropy
fBU , for any x X. In particular, A = B, if A B and A B. INS (A, B) between A and B should satisfy the following condi-
The complement set of A, denoted by Ac , is defined as Ac = (1) INS (A, B) = INS (B, A);
{(x, [ fAL , f U ], [iLA , iU ], [tAL , t U ])|x X}. (2) INS (A, B) = INS (Ac , Bc ), where Ac and Bc are the com-
plement sets of A and B, respectively, as defined in Defi-
3. A transformation operator and a cross-entropy nition 5; and
measure of SNNs (3) INS (A, B) 0 and INS (A, B) = 0, if and only if A = B.
In this section, a transformation operator is developed to con- Definition 10 Let A = (tA , iA , fA ) and B = (tB , iB , fB ) be two
vert INNs into SNNs. Moreover, a simplified neutrosophic cross- SNNs, and then the cross-entropy of A and B can be defined as
entropy measure is defined. follows:

3.1. A transformation operator between INNs and SNNs 2tA 2iA 2 fA

INS (A, B) = tA ln + iA ln + fA ln . (1)
Bustince and Burillo (1995) proposed an operator Hp,q , which tA + tB iA + iB fA + fB
can transform each IVFS into an IFS. As an improved extension
of Hp,q , an operator Hp,q,r is defined for converting each INN Equation (1) can indicate the degree of discrimination of
into an SNN. A from B. It is obvious that INS (A, B) is not symmetrical with
4 Z.-p. TIAN ET AL.

respect to its arguments. Therefore, a modified symmetric dis- information of each alternative ai (i = 1, 2, . . . , m) on the
crimination information measure based on INS (A, B) can be criterion c j ( j = 1, 2, . . . , n) in the form of INNs.
defined as In general, there are two types of criterion, namely maximis-
ing and minimising criteria. In order to make the criterion types
DNS (A, B) = INS (A, B) + INS (B, A). (2) constant, the minimising criteria need to be transformed into
maximising ones. Suppose the standardised matrix is expressed
The larger DNS (A, B) is, the larger the difference between A as R = [ri j ]. The original decision matrix B can then be con-
and B will be, and vice versa. verted into R based on the primary transformation principle of
Xu and Hu (2010), where
Proposition 1: The measures defined in Equations (1) and (2) are
the simplified neutrosophic cross-entropy, and satisfy conditions      

bi j = tiLj , tiUj , iLij , iUij , fiLj , fiUj ,
(1)(3) given in Definition 9.

Proof: Clearly, conditions (1) and (2) are obvious. The proof ri j =
for maximising criterion c j
 L U   L U   L U  , (3)
of condition (3) is shown below.
bi j = fi j , fi j , ii j , ii j , ti j , ti j ,

for minimising criterion c j
Consider the function f (x) = x ln x, where x (0, 1].
Then, f (x) = 1 + ln x and f (x) = 1/x > 0, where x (0, 1].
Accordingly, f (x) = x ln x is a convex function. Therefore, in which bci j is the complement set of bi j , defined in Definition 7.
for any two points x1 , x2 (0, 1], the inequality f (x1 )+2 f (x2 )
f ( x1 +x
) holds. 4.1. A fuzzy cross-entropy based on TOPSIS
Utilise f (x) = x ln x in the above inequality and x1 ln x1 + The absolute positive ideal solution (PIS) and the absolute nega-
x2 ln x2 (x1 + x2 ) ln x1 +x
0 can be obtained; in this case, tive ideal solution (NIS) of INSs are, respectively, denoted by a+
the equality holds only if x1 = x2 . Similarly, the following equa- and a , and can be expressed as follows (Chi & Liu, 2013):
tion can be obtained:
a+ = ([1, 1], [0, 0], [0, 0]) and a = ([0, 0], [1, 1], [1, 1]) .
DNS (A, B) = INS (A, B) + INS (B, A)
tA +tB
= (tA ln tA +tB ln tB ) (tA +tB ) ln In order to obtain the cross-entropy or degree of discrimina-
tion between ai (i = 1, 2, . . . , m) and the ideal solutions, each
INN is transformed into an SNN based on the operator Hp,q,r ,
iA +iB
+ (iA ln iA +iB ln iB ) (iA +iB ) ln which was introduced in Definition 8. Let pi j , qi j , ri j [0, 1],
  2  and then any INN, denoted by ([tiLj , tiUj ], [iLij , iUij ], [ fiLj , fiUj ]), can
I be transformed into the following form:
fA + fB
+ ( fA ln fA + fB ln fB ) ( fA + fB ) ln . 
  2  Hp,q,r (ri j ) = tiLj + pi jWti j , iLij + qi jWii j , fiLj + ri jW fi j , (4)

Because T 0, I 0 and F 0, DNS (A, B) 0 holds; where Wti j = tiUj tiLj , Wii j = iUij iLij and W fi j = fiUj fiLj .
DNS (A, B) = 0 holds only if tA = tB , iA = iB and fA = fB , Using Equations (1), (2) and (4), the degree of discrimination
namelyA = B. of ai (i = 1, 2, . . . , m) from the ideal solutions a+ and a_ with
respect to c j ( j = 1, 2, . . . , n) can be, respectively, calculated as
Example 2: Assume two SNNs A = (0.8, 0.1, 0.2) and B = follows:
(0.5, 0.3, 0.4). Then, the following result can be obtained by
applying Equations (1) and (2):
2(tiLj + pi jWti j )
i j = (ti j + pi jWti j )ln
DNS (A, B) = INS (A, B) + INS (B, A) = 0.1212. (tiLj + pi jWti j + 1)
+ ln + (iLij + qi jWii j )ln2
4. An MCDM approach based on the cross-entropy (tiLj + pi jWti j + 1)
+ ( fiLj + ri jW fi j )ln2, (5)
This section presents an approach that is based on the cross-
entropy and TOPSIS for solving interval neutrosophic MCDM 2(iLij + qi jWii j )
problems with incomplete weight information. D
i j = (ti j + pi jWti j ) ln 2 + (ii j + qi jWii j ) ln

For an MCDM problem, let A = {a1 , a2 , . . . , am } (iLij + qi jWii j + 1)

be a set consisting of m alternatives, and let C = 2
+ ln
{c1 , c2 , . . . , cn } be a set consisting of n criteria. Assume (iLij + qi jWii j + 1)
that w = (w1 , w2 , . . . , wn ) is the weight vector of crite-
n 2( fiLj + ri jW fi j )
ria, where w j [0, 1] and j=1 w j = 1. Let B = [bi j ] = + ( fiLj + ri jW fi j ) ln
( fiLj + ri jW fi j + 1)
[([tiLj , tiUj ], [iLij , iUij ], [ fiLj , fiUj ])]mn be the decision matrix, where
[bi j ] = [([tiLj , tiUj ], [iLij , iUij ], [ fiLj , fiUj ])]mn is the evaluation + ln . (6)
( fiLj + ri jW fi j + 1)

In the TOPSIS method, the closeness coefficient tion with respect to qi j or ri j . Thus, D+i j can reach its max-
D i j = D +
i j /(Di j + Di j ) denotes the performance of ai (i = 1, 2, imum and D can reach the minimum if pi j = 0 and qi j =
. . . , m) on c j ( j = 1, 2, , n). Then, the performance of ai , ri j = 1. Likewise, D+ reaches its minimum and D
ij i j reaches
denoted by Di , can be obtained by using Equations (5) and (6): the maximum if pi j = 1 and qi j = ri j = 0. As a result, it is
easy to understand that Di j reaches its minimum when pi j = 0

Di = w j Di j = wj
, (7) and qi j = ri j = 1, and reaches its maximum when pi j = 1 and
j=1 j=1
D +
i j + Di j qi j = ri j = 0.
Then, DLij and DUij can be obtained by integrating Equa-
where w j represents the weight of c j . Therefore, the larger Di is, tions (5) and (6) into Di j , respectively:
the better ai will be.

2iUij 2 fiUj
tiLj ln 2 + iUij ln (ii j +1)
U + fiUj ln ( fiUj +1)
+ ln 2
(iUij +1)
+ ln 2
( fiUj +1)
DLij = 2tiLj 2iUij 2 fiUj
, (8)
(tiLj + iUij + fiUj ) ln 2 + tiLj ln (tiLj +1)
+ iUij ln (iUij +1)
+ fiUj ln ( fiUj +1)
+ ln 2
(tiLj +1)
+ ln 2
(iUij +1)
+ ln 2
( fiUj +1)
2iLij 2 fiLj
tiUj ln 2 + iLij ln (iLij +1)
+ fiLj ln ( fiLj +1)
+ ln 2
(iLij +1)
+ ln 2
( fiLj +1)
DUij = 2tiUj 2iLij 2 fiLj
, (9)
(tiUj + iLij + fiLj ) ln 2 + tiUj ln (tiUj +1)
+ iLij ln (iLij +1)
+ fiLj ln ( fiLj +1)
+ ln 2
(tiUj +1)
+ ln 2
(iLij +1)
+ ln 2
( fiLj +1)

where DLij and DUij are also the minimum and maximum of Di j =
In view of the fact that an INS is characterised by a truth-
D +
i j /(Di j + Di j ), respectively.
membership function, a indeterminacy-membership function
Based on the analysis above, Equation (7) can be rewritten
and a falsity-membership function, whose values are intervals
rather than specific numbers, it is infeasible to designate an SNN
for the given INN by artificially choosing only certain pi j , qi j and  

ri j in Equation (4) to indicate the evaluation information. This Di = w j Di j = w j DLij , DUij , (10)
is because it may lead to the distortion or loss of the original j=1 j=1
information (Zhang & Yu, 2012).
By combining Equations (5) and (6), Di j becomes a compos-
where Di D j means that ai (i = 1, 2, . . . , m) is not inferior to
ite function and may take different values as the numerical vari-
a j ( j = 1, 2, . . . , m), and the weight information is incompletely
ations pi j , qi j and ri j change. In order to avoid information loss,
known. Obviously, Di represents the comprehensive evaluation
an interval Di j = [DLij , DUij ] is applied to represent the perfor-
values, and stands for the preference of ai ; that is, the larger Di
mance of ai (i = 1, 2, . . . , m) on c j ( j = 1, 2, . . . , n), where DLij is, the better ai will be.
and DUij are the lower and upper bounds, respectively.
In the following paragraphs, DLij and DUij are determined
4.2. Fuzzy linear programming models for determining
through mathematical derivation.
criterion weights
As shown in Equations (5) and (6), both D+
i j and Di j are mul- In the decision-making process, the importance of different cri-
tivariate continuous functions with respect to pi j , qi j and ri j . teria should be taken into consideration. Suppose 0 denotes the
They can also reach the maximum and minimum in the domain set of all the weight vectors, and
pi j , qi j , ri j [0, 1]. Then, calculate the partial derivative of D+ij

and D i j on pi j , qi j and ri j : 

0 = (w1 , w2 , . . . , wn ) w j 0( j = 1, 2, . . . , n), wj = 1 .

D+ D+
ij 2(tiLj + pi jWti j ) ij
= Wti j ln , = Wii j ln 2 0, and
pi j (tiLj + pi jWti j + 1) qi j
In some actual decision-making situations, the incomplete
ij information regarding the criterion weights provided by DMs
= W fi j ln 2 0. can usually be constructed using several basic ranking forms
ri j
(Dubey et al., 2012; Li, 2011). These weight information struc-
tures may be expressed in the following five basic relations,
Since 2(tiLj + pi jWti j ) tiLj + pi jWti j + 1, ij
pi j
0. which are denoted by the subsets s (s = 1, 2, . . . , 5) in 0 ,
D D 2(iLij +qi j Wii j ) respectively (Chen, 2014).
Similarly, ij
pi j
= Wti j ln 2 0, ij
qi j
= Wii j ln (iL +qi jWi +1)
(1) A weak ranking:
ij ij
D 2( fiLj +ri j W fi j )
0, and ij
ri j
= W fi j ln ( f L +ri jW f +1)
ij ij
1 = (w1 , w2 , . . . , wn ) 0 w j1 w j2 forall j1 1 and j2 1 ,
This means that D+
is a monotone decreasing function
with respect to pi j and a monotone increasing function with
respect to qi j or ri j . Moreover, D where 1 and 1 are two disjoint subsets of the subscript index
i j is a monotone increasing
set N = {1, 2, . . . , n} of all criteria.
function with respect to pi j and a monotone decreasing func-
6 Z.-p. TIAN ET AL.

(2) A strict ranking: case of inconsistent preference information:

2 = {(w1 , w2 , . . . , wn ) 0 |w j1 w j2 j1 j2 for all max D = w j DM
j1 2 and j2 2 }, i=1 j=1
min E = e +
(1) j1 j2 + e(2) j1 j2 + e(3) j1 j2 j3 j4 + e(4) j1 + e(4) j1 + e(5) j1 j2
j1 , j2 , j3 , j4 N

where j1 j2 > 0 is a constant, and 2 and 2 are two disjoint sub-
(w1 , w2 , . . . , wn ) 

sets of N.
e(1) j1 j2 0 j1 1 and j2 1

(3) A ranking of differences: e
(2) j1 j2 0 j1 2 and j2 2

e (3) j1 j2 j3 j4 0 j1 3 , j2 3 , j3 3 and j4 3

3 = {(w1 , w2 , . . . , wn ) 0 |w j1 w j2 w j3 w j4 for all

e (4) j1 0, e(4) j1 0 j1 4

e(5) j1 j2 0 j1 5 and j2 5
j1 3 , j2 3 , j3 3 and j4 3 },
where DM i j = (Di j + Di j )/2. By solving Model (M2), the optimal
where 3 , 3 , 3 and 3 are four disjoint subsets of N. weight vector w = (w1 , w2 , , wn ) and the optimal deviation
(4) An interval form: values e +
(1) j1 j2 , e(2) j1 j2 , e(3) j1 j2 j3 j4 , e(4) j1 , e(4) j1 and e(5) j1 j2 can be
4 = (w1 , w2 , . . . , wn ) 0  j1 w j1 j1 for all j1 4 , Considering an MCDM problem that contains incomplete
and consistent preference information, Model (M1) can be
where j1 > 0 and j1 > 0 are constants, satisfying j1 > j1 , applied to obtain the best weight vector, and the optimisation
and 4 is a subset of N. model can be easily solved by using the simplex method. For
(5) A ranking with multiples: an MCDM problem that contains incomplete and inconsistent
preference information, Model (M2) can be employed to deter-
5 = {(w1 , w2 , . . . , wn ) 0 |w j1 j1 j2 w j2 for all mine the optimal solution.
j1 5 and j2 5 },
4.3. The proposed algorithm with incomplete criteria
where j1 j2 > 0 is a constant, and 5 and 5 are two disjoint sub-
In view of the determination of the weight vector of criteria, the
sets of N.
comprehensive preference of each alternative ai can be denoted
Let  denote a set of the known information on the crite-
nan interval value;n that is, Equation (10) is revised as Di =

rion weights, and  = 1 2 3 4 5 . Given the con- n
j=1 w j Di j = [ j=1 w j Di j , j=1 w j Di j ] by using the opera-
ditions in , the optimal weight values of the criteria in Equation
tional laws of interval values given in Definition 1. Then, a
(10) can be determined via the following linear programming
pairwise comparison must be made between the alternatives,
and subsequently the pairwise comparison matrix (likelihood
  matrix) P can be constructed as follows:
max D = ni=1 nj=1 w j DM ij
, (M1)
s.t. (w1 , w2 , . . . , wn )  p11 p12 p1m
 p21 p22 p2m

where DM i j = (Di j + Di j )/2.
L U P = pi j mm = . .. .. .. , (11)
.. . . .
The DMs might express inconsistent opinions about the pref-
erences and weight information in the case of contingency. pm1 pm2 pmm
Under these circumstances, a multi-objective nonlinear pro-
gramming model using a goal programming technique can be where pi j = p(ai a j ) = p(Di D j ) = max
used to tackle problems that involve inconsistent weight infor- {1 max{ L(D )+L(D ) , 0}, 0}, and L(Di ) = DUi DLi .
mation. For j1 = j2 = j3 = j4 ,  is revised as  by introducing i j
According to Xu and Da (2003), the ranking vector of the
several non-negative deviation variables: likelihood matrix can be defined as follows:
 = {(w1 , w2 , . . . , wn ) 0 |w j1 + e (1) j1 j2 w j2 for all j=1 pi j + 2 1
i = , i = 1, 2, . . . , m. (12)
j1 1 and j2 1 ; m(m 1)
w j1 w j2 + e (2) j1 j2 j1 j2 for all j1 2 and j2 2 ;
Consequently, the ranking of all alternatives can be obtained
w j1 w j2 w j3 + w j4 + e (3) j1 j2 j3 j4 0 for all according to the descending order of i (i = 1, 2, , m). That
j1 3 , j2 3 , j3 3 and j4 3 ; is, the larger i is, the better the alternative ai will be.
w j1 + e + Based on the above analysis, an approach can be developed
(4) j1 j1 ,w j1 e(4) j1 j1 for all j1 4 ;
for an MCDM problem that contains three key stages: (1) col-
w j1 /w j2 + e
(5) j1 j2 j1 j2 for all j1 5 and j2 5 }. lection and normalisation stage, (2) determination stage and (3)
selection stage. A conceptual model of the proposed approach is
Furthermore, based on Model (M1), the following bi- shown in Figure 1.
objective nonlinear programming model can be established in The main steps are outlined as follows.

Collection and normalisation stage (Step 1)

Evaluation values
Decision-makers Collect the criterion weight information
and normalise the decision matrix

Normalised evaluation values

Determination stage (Steps 2-4)

Calculate the lower Interval values Construct the programming model and
Ranking and upper bounds identify the optimal weight vector
Selection stage (Step 5)
Interval values Optimal weights
Rank all alternatives

Calculate the comprehensive performance and obtain the

ranking vector (Using the likelihood-based comparison method)

Figure . A ow chart of the proposed approach.

Step 1: Normalise the decision matrix. metals. It is also the largest manufacturer of multi-species non-
Use Equation (3) to transform B into R. For convenience, the ferrous metals, with the exception of aluminium, in China. To
normalised values of ai (i = 1, 2, . . . , m) with respect to c j ( j = expand its main business, the company regularly engages in
1, 2, . . . , n) are also expressed as ([tiLj , tiUj ], [iLij , iUij ], [ fiLj , fiUj ]). overseas investment and a department consisting of executive
Step 2: Calculate the lower and upper bounds of Di j . managers and several experts in the field has been established to
Use Equations (8) and (9) to derive the lower bound DLij and make decisions regarding global mineral investment. This over-
the upper bound DUij , respectively. seas investment department recently decided to select a pool
Step 3: Identify the optimal weight vector and calculate the of alternatives from several foreign countries based on prelimi-
preference of each alternative. nary surveys. After a thorough investigation, five countries were
Solve Model (M1) or (M2) to identify the optimal weight taken into consideration, denoted by {a1 , a2 , , a5 }. There are
vector, and calculate the comprehensive performance Di (i = many factors that affect the investment environment, but four
1, 2, . . . , m) by using the operational laws of the interval values were chosen based on the experience of the departments per-
in Definition 1. sonnel, namely c1 : resources; c2 : politics and policy; c3 : econ-
Step 4: Construct the likelihood matrix and obtain the rank- omy; and c4 : infrastructure.
ing vector. The members of the overseas investment department
Construct the likelihood matrix P by using Equation (11) have met to determine the evaluation information. Con-
and obtain the ranking vector = (1 , 2 , . . . , m ) based on sequently, following a heated discussion, they came to a
Equation (12). consensus on the final evaluations which were expressed
Step 5: Determine the ranking of all alternatives. by INNs shown in Table 1 Moreover, they were only able
Determine the ranking of all alternatives according to the to provide incomplete information on the weights, that is,
descending order of i = (1, 2, . . . , m) and select the optimal  = {0.15 w1 0.3, 0.15 w2 0.25, 0.25 w3 0.4,
one(s). 0.3 w4 0.45, 2.5w1 w3 }.

5. An illustrative example 5.1. Illustration of the proposed approach

In this section, the example of an investment appraisal project In the following steps, the main procedures of obtaining the
is used to demonstrate the application of the proposed MCDM optimal ranking of alternatives are presented.
approach; then its validity and effectiveness will be tested Step 1: Normalise the decision matrix.
through a comparative analysis. As all the criteria are maximising type, the matrix does not
The following case is adapted from Wang et al. (2015). need to be normalised, i.e., R = B.
ABC Nonferrous Metals Co. Ltd. is a large state-owned com- Step 2: Calculate the lower and upper bounds of Di j .
pany whose main business is the deep processing of non-ferrous Derive the lower bound DLij and the upper bound DUij by using
Equations (8) and,(9), respectively:

[0.5525, 0.7141] [0.5363, 0.6410] [0.6130, 0.7571] [0.6726, 0.7613]
  [0.4966, 0.7571] [0.4964, 0.7853] [0.5363, 0.6429] [0.5297, 0.6962]

Di j =
[0.5610, 0.7528] [0.4462, 0.6148] [0.5525, 0.6753] [0.4568, 0.5965] .

54 [0.3580, 0.4433] [0.4433, 0.5825] [0.4350, 0.5525] [0.6962, 0.8171]
[0.4406, 0.5637] [0.6962, 0.8171] [0.5873, 0.7141] [0.4964, 0.6272]
8 Z.-p. TIAN ET AL.

Table . The evaluation information.

c1 c2 c3 c4

a1 ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.], [.,.])

a2 ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.], [.,.])
a3 ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.], [.,.])
a4 ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.], [.,.])
a5 ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.],[.,.]) ([.,.],[.,.], [.,.])

Step 3: Identify the optimal weight vector and calculate the 5.2. Comparative analysis and discussion
preference of each alternative. In order to further verify the feasibility and effectiveness of the
Because no inconsistent weight information exists in the proposed approach, a comparative analysis is now conducted
evaluation, Model (M1) can be applied to identify the optimal using six existing methods with the analysis being based on the
weight vector: same illustrative example.
(1) Chi and Lius method (2013) contains two major phases
max D = 2.8198w1 + 3.0295w2 + 3.033w3 + 3.175w4 (criterion weights determination and ranking obtain-
0.15 w1 0.3 ment with TOPSIS). First, the maximising deviation

w2 0.24 method is developed to determine the criterion weights

0.25 w 0.4 and then an extended TOPSIS method is employed to
s.t. rank the alternatives.

0.3 w 4 0.45

w (2) In Yes method (2014b), the distance-based similarity

1 3
w = 1 measures are employed, which involves aggregating the
j=1 j
weighted similarity measures between each alternative
and the PIS.
The optimal weight vector can be obtained as w = (3) In Zhang et al.s methods (2014), first, the comprehensive
(0.15, 0.15, 0.375, 0.325). Then, Di (i = 1, 2, , 5) can be INNs are aggregated by using the INNWA or INNWG
calculated by referring to the operational laws of interval values operators, and then the ranking vectors can be obtained
in Definition 1: by constructing the likelihood matrices based on the
score function. Moreover, Zhang et al. (2015b) devel-
D1 = [0.6118, 0.7346], D2 = [0.5222, 0.6987], oped an outranking method for MCDM on the basis of
D3 = [0.5067, 0.6522], the score function constructing the outranking relation
matrix. Since Zhang et al.s method (2015b) does not take
the criterion weights into consideration but the illustra-
tive example does, it is necessary to construct the out-
ranking relation matrix after calculating the weighted
D4 = [0.5096, 0.6266] and D5 = [0.5521, 0.6788]. evaluation values.
Considering the criterion weights obtained using the pro-
Step 4: Construct the likelihood matrix and obtain the rank- posed programming model, the results obtained by different
ing vector. methods are summarised in Table 2.
Use Equation (11) to construct the likelihood matrix P It can be seen that there are some differences between them.
and obtain the ranking vector = (1 , 2 , . . . , 5 ) based on The reasons for the inconsistency of the rankings are explained
Equation (12): as follows.
(1) The difference in the ranking results of the proposed
 approach and that of Chi and Liu (2013) is the sequence
P = pi j 55

0.5 0.7096 0.8493 0.9384 0.7316 Table . The ranking results of the dierent methods.
0.2904 0.5 0.5962 0.6444 0.4836
Methods Ranking results
0.1507 0.4038 0.5 0.5434 0.3679 ,

0.0616 0.3556 0.4566 0.5 0.3057 Chi and Lius method () a1 a5 a2 a4 a3
Yes methods (b)
0.2684 0.5164 0.6321 0.6943 0.5 Similarity measure based on the Hamming a1 a5 a2 a4 a3
1 = 0.2614, 2 = 0.2007, 3 = 0.1733, distance
Similarity measure based on the Euclidean a1 a5 a2 a3 a4
4 = 0.1590 and 5 = 0.2056. distance
Zhang et al.s methods ()
Method based on the INNWA operator a1 a5 a2 a4 a3
Step 5: Determine the ranking of all alternatives. Method based on the INNWG operator a1 a5 a2 a3 a4
According to the descending order of i = (1, 2, . . . , 5), the Zhang et al.s method (b) (p = 0.2 and a1 a2 {a3 , a4 , a5 }
q = 0.1)
ranking of all alternatives is a1 a5 a2 a3 a4 and the The proposed approach a1 a5 a2 a3 a4
best one is a1 .

Table . The ranking results obtained by revising the methods of Ye (b). when manually providing the parameters p and q, it is
Methods Ranking results difficult to avoid subjective randomness. Therefore, the
result obtained by the outranking method is not always
Similarity measure based on the Hamming a1 a5 a2 a3 a4 reliable.
Similarity measure based on the Euclidean a1 a5 a2 a3 a4 According to the comparative analysis, the following advan-
distance tages over the other methods can be outlined.
(1) The calculations required for the proposed approach are
relatively straightforward and time-saving, and the bur-
of a3 and a4 . In Chi and Lius method (2013), the rela- den of computation can be greatly decreased with the
tive closeness coefficients (performance of each alterna- help of the proven mathematical derivation.
tive) are conducted based on the relative ideal solutions, (2) In the proposed approach, the interval closeness coeffi-
which have certain drawbacks. First, it is not easy to cients are conducted to rank alternatives. In this way, the
choose the appropriate PIS and NIS with INNs because fuzziness of the original information can be maintained
each INN has three interval elements. Second, the PIS and fully utilised. Therefore, the proposed approach is
and NIS are closely related to the number of alterna- more competent in interval neutrosophic MCDM than
tives as well as the evaluation values. Thus, they may vary the other methods considered.
as the original information changes. Third, the relative (3) In the proposed approach, the transformation opera-
closeness coefficients are in the form of crisp real num- tor is employed to convert INNs into SNNs, which can
bers, which may cause information loss and affect the avoid various kinds of aggregation operators process-
ranking results. Finally, suppose that there are m alter- ing directly with INNs. Furthermore, the parameters
natives and n criteria to be evaluated. In order to deter- of the transformation operator are determined through
mine the criterion weights, Chi and Lius method (2013) mathematical derivation and not artificially produced.
needs to derive m m n distance measures, whereas Thereby, the final ranking obtained by the proposed
the proposed approach only needs to calculate m n approach is more conclusive than those produced by
cross-entropy measures. Thus, it takes less time than that the other methods, and it is evident that the proposed
of Chi and Liu (2013). approach is accurate and reliable.
(2) Similarly, the order of a3 and a4 is the only difference
between the proposed approach and the first method of
Ye (2014b). This is because the comparison methods in 6. Conclusions
Ye (2014b) only consider the weighted similarity mea- INSs are flexible at expressing the uncertain, imprecise, incom-
sures between each alternative and the PIS. If only the PIS plete and inconsistent information that is very common in scien-
is taken into account and the NIS is ignored, the rank- tific and engineering situations; therefore, the study of MCDM
ing of alternatives may be incorrectly reversed and this methods with INSs is highly significant. In this paper, a trans-
may be amplified in the final results. However, the rank- formation operator and cross-entropy were defined. Conse-
ing result in this method will be identical to that of the quently, an MCDM method was established based on cross-
proposed approach in a situation where, simultaneously, entropy and TOPSIS, which calculated the cross-entropy after
the PIS is replaced with the absolute one and the abso- transforming INNs into SNNs on the basis of the transforma-
lute NIS is not ignored; moreover, the closeness coeffi- tion operator. Furthermore, it aggregated the performances of
cient would have to be utilised to determine the ranking alternatives into interval numbers, from which two mathemati-
of alternatives. The updated results are shown in Table 3. cal programming models were constructed to identify the crite-
Therefore, the methods in Ye (2014b) are not reliable rion weights. Finally, a ranking result was obtained by compar-
enough. ing these weighted interval numbers with a possibility degree
(3) The positions of a3 and a4 obtained by the method based method.
on the INNWA operator (Zhang et al., 2014) are not The advantages of this study are that the approach is both
consistent with either those obtained by the method simple and convenient to compute and effective at decreasing
based on the INNWG operator (Zhang et al., 2014) the loss of evaluation information. The feasibility and validity of
or the proposed approach. This may be caused by the the proposed approach have been verified through the illustra-
inherent characteristics of aggregation operators, as the tive example and comparative analysis. The comparison results
INNWA operator focuses on the impact of the overall demonstrated that the proposed approach can provide more
criterion values, while the INNWG operator emphasises reliable and precise outcomes than other methods. Therefore,
the impact of a single item. Additionally, the outrank- this approach has great application potential in solving MCDM
ing method of Zhang et al. (2015b) can only yield partial problems in an interval neutrosophic environment, in which cri-
orders of alternatives, in which a3 , a4 and a5 are indis- terion values with respect to alternatives are evaluated by the
tinguishable. This method has to convert the INNs into form of INNs and the criterion weights are incomplete.
real numbers and artificially set both the threshold p and
indifference threshold q before constructing the domi-
nance relations. It is by nature inappropriate to replace Acknowledgements
the INNs with real numbers, and it may lead to infor- The authors thank the editors and anonymous reviewers for their helpful
mation loss in the transformation process. Furthermore, comments and suggestions.
10 Z.-p. TIAN ET AL.

Disclosure statement for green supplier selection. Journal of Intelligent & Fuzzy Systems,
No potential conflict of interest was reported by the authors. 28(1), 117126.
Chen, S.M., & Chang, C.H. (2015). A novel similarity measure between
Atanassovs intuitionistic fuzzy sets based on transformation tech-
niques with applications to pattern recognition. Information Sciences,
This work was supported by the National Natural Science Foundation of 291, 96114.
China [grant number 71571193], [grant number 71431006]. Chen, S.M., Cheng, S.H., & Chiou, C.H. (2016). Fuzzy multi-attribute group
decision making based on intuitionistic fuzzy sets and evidential rea-
Notes on contributors soning methodology. Information Fusion, 27, 215227.
Zhang-peng Tian received his BS degree in electronic commerce from Tian- Chen, T.Y. (2014). Interval-valued intuitionistic fuzzy QUALIFLEX method
jin Chengjian University, Tianjin, China, in 2012. He is currently working with a likelihood-based comparison approach for multiple criteria
towards the MS degree with the Business School, Central South University, decision analysis. Information Sciences, 261, 149169.
Changsha, China. Chi, P.P., & Liu, P.D. (2013). An extended TOPSIS method for the multi-
ple attribute decision-making problems based on interval neutrosophic
Hong-yu Zhang received her MS degree in computer software and the- set. Neutrosophic Sets and Systems, 1, 6370.
ory and PhD degree in management science and engineering from Central Deli, I., Ali, M., & Smarandache, F. (2015). Bipolar neutrosophic sets and
South University, Changsha, China, in 2005 and 2009, respectively. She is their application based on multi-criteria decision making problems.
currently an associate professor with the Business School, Central South Retrieved from
University. Her research interests include information management and Dubey, D., Chandra, S., & Mehra, A. (2012). Fuzzy linear programming
its applications in production operations. Her current research focuses on under interval uncertainty based on IFS representation. Fuzzy Sets and
remanufacturing production management and decision-making theory. Systems, 188(1), 6887.
Jing Wang received her MSc degree in information engineering from the Guo, Y.H., & Sengr, A. (2014). A novel image edge detection algorithm
University of Osnabrueck, Osnabrueck, Germany, in 2006. She is currently based on neutrosophic set. Computers & Electrical Engineering, 40(8),
working towards the PhD degree with the Business School, Central South 325.
University, Changsha, China. She is also a lecturer with International Col- Hwang, C.L., & Yoon, K. (1981), Multiple attribute decision-making and
lege, Central South University of Forestry and Technology. Her current applications. Berlin: Springer-Verlag.
research interests include decision-making theory and application, risk Li, D.F. (2011). Closeness coefficient based nonlinear programming method
management and control, and information management. for interval-valued intuitionistic fuzzy multi-attribute decision-making
with incomplete preference information. Applied Soft Computing,
Jian-qiang Wang received his PhD degree in management science and 11(4), 34023418.
engineering from Central South University, Changsha, China, in 2005. He Liu, B.S., Shen, Y.H., Zhang, W., Chen, X.H., & Wang, X.Q. (2015).
is currently a professor with the Business School, Central South University. An interval-valued intuitionistic fuzzy principal component anal-
His current research interests include decision-making theory and applica- ysis model-based method for complex multi-attribute large-group
tion, risk management and control, and information management. decision-making. European Journal of Operational Research, 245(1),
Xiao-hong Chen received her PhD degree in management from the Tokyo 209225.
University of Technology, Tokyo, Japan, in 1999. She is currently a profes- Liu, P.D., & Wang, Y.M. (2015). Interval neutrosophic prioritized OWA
sor with the Business School, Central South University, Changsha, China. operator and its application to multiple attribute decision making. Jour-
Her current research interests include decision theory and methods, deci- nal of Systems Science and Complexity. Advance online publication. doi:
sion support systems, resource-saving, and the environment-friendly soci- 10.1007/s11424-015-4010-7
ety. She has published academic papers in many journals, including Decision Macedo, P., & Scotto, M. (2014). Cross-entropy estimation in technical effi-
Support System and Expert Systems with Applications. ciency analysis. Journal of Mathematical Economics, 54, 124130.
Majumdar, P., & Samant, S.K. (2014). On Similarity and entropy of neutro-
sophic sets. Journal of Intelligent & Fuzzy Systems, 26(3), 12451252.
ORCID Meng, F.Y., & Chen, X.H. (2015). Entropy and similarity measure for Atan-
Zhang-peng Tian nasovs interval-valued intuitionistic fuzzy sets and their application.
Hong-yu Zhang Fuzzy Optimization Decision-making. Advance online publication. doi:
Jing Wang 10.1007/s10700-015-9215-7
Jian-qiang Wang Mohan, J., Krishnaveni, V., & Guo, Y.H. (2013). MRI denoising using non-
Xiao-hong Chen local neutrosophic set approach of wiener filtering. Biomedical Signal
Processing and Control, 8(6), 779791.
Peng, J.J., Wang, J.Q., Wang, J., Zhang, H.Y., & Chen, X.H. (2015). Simpli-
References fied neutrosophic sets and their applications in multi-criteria group
Atanassov, K.T. (1986). Intuitionistic fuzzy sets. Fuzzy Sets and Systems, decision-making problems. International Journal of Systems Science.
20(1), 8796. Advance online publication. doi:10.1080/00207721.2014.994050.
Atanassov, K.T., & Gargov, G. (1989). Interval valued intuitionistic fuzzy Peng, J.J., Wang, J.Q., Wu, X.H., Wang, J., & Chen, X.H. (2015). Multi-
sets. Fuzzy Sets and Systems, 31(3), 343349. valued neutrosophic sets and power aggregation operators with their
Bellman, R., & Zadeh, L.A. (1970). Decision-making in A fuzzy environ- applications in multi-criteria group decision-making problems. Inter-
ment. Management Science, 17(4), 141164. national Journal of Computational Intelligence Systems, 8(2), 345
Broumi, S., & Smarandache, F. (2014a). Correlation coefficient of interval 363.
neutrosophic set. Applied Mechanics and Materials, 436, 511517. Peng, J.J., Wang, J.Q., Wu, X.H., Zhang, H.Y., & Chen, X.H. (2014). The
Broumi, S., & Smarandache, F. (2014b). Cosine similarity measure of inter- fuzzy cross-entropy for intuitionistic hesitant fuzzy sets and their appli-
val valued neutrosophic sets. Neutrosophic Sets and Systems, 5, 1520. cation in multi-criteria decision-making. International Journal of Sys-
Bustince, H., & Burillo, P. (1995). Correlation of interval-valued intuition- tems Science, 46(13), 23352350.
istic fuzzy sets. Fuzzy Sets and Systems, 74(2), 237244. Peng, J.J., Wang, J.Q., Zhang, H.Y., & Chen, X.H. (2014). An outranking
Caballero, R., Hernndez-Daz, A.G., Laguna, M., & Molina, J. (2015). approach for multi-criteria decision-making problems with simplified
Cross entropy for multi-objective combinatorial optimization prob- neutrosophic sets. Applied Soft Computing, 25, 336346.
lems with linear relaxations. European Journal of Operational Research, Rivieccio, U. (2008). Neutrosophic logics: Prospects and problems. Fuzzy
243(3), 362368. Sets and Systems, 159(14), 18601868.
Cao, Q.W., Wu, J., & Liang, C.Y. (2015). An intuitionistic fuzzy judgement Sengupta, A., & Pal, T.K. (2000). On comparing interval numbers. European
matrix and TOPSIS integrated multi-criteria decision-making method Journal of Operational Research, 127(1), 2843.

Shang, X.G., & Jiang, W.S. (1997). A note on fuzzy information measures. Xu, Z.S., & Hu, H. (2010). Projection models for intuitionistic fuzzy multiple
Pattern Recognition Letters, 18(5), 425432. attribute decision-making. International Journal of Information Tech-
Smarandache, F. (1999), A unifying field in logics. Neutrosophy: Neutrosophic nology & Decision-Making, 9(2), 267280.
probability, set and logic. Rehoboth, MA: American Research Press. Ye, J. (2011). Fuzzy cross entropy of interval-valued intuitionistic fuzzy sets
Solis, A.R., & Panoutsos, G. (2013). Granular computing neural-fuzzy mod- and its optimal decision-making method based on the weights of alter-
elling: A neutrosophic approach. Applied Soft Computing, 13(9), 4010 natives. Expert Systems with Applications, 38(5), 61796183.
4021. Ye, J. (2013). Multicriteria decision-making method using the correlation
Sun, H.X., Yang, H.X., Wu, J.Z., & Yao, O.Y. (2015). Interval neutrosophic coefficient under single-value neutrosophic environment. International
numbers choquet integral operator for multi-criteria decision making. Journal of General Systems, 42(4), 386394.
Journal of Intelligent & Fuzzy Systems, 28(6), 24432455. Ye, J. (2014a). A multicriteria decision-making method using aggregation
Tan, C.Q., Ma, B. J., & Wu, D.D., & Chen, X.H. (2014). Multi-criteria operators for simplified neutrosophic sets. Journal of Intelligent & Fuzzy
decision-making methods based on interval-valued intuitionistic fuzzy Systems, 26(5), 24592466.
sets. International Journal of Uncertainty, Fuzziness and Knowledge- Ye, J. (2014b). Similarity measures between interval neutrosophic sets and
Based Systems, 22(3), 475494. their applications in multicriteria decision-making. Journal of Intelli-
Vlachos, I.K., & Sergiadis, G.D. (2007). Intuitionistic fuzzy information gent & Fuzzy Systems, 26(1), 165172.
applications to pattern recognition. Pattern Recognition Letters, 28(2), Ye, J. (2014c). Single valued neutrosophic cross-entropy for multicrite-
197206. ria decision-making problems. Applied Mathematical Modelling, 38(3),
Wan, S.P., & Dong, J.Y. (2014). A possibility degree method for interval- 11701175.
valued intuitionistic fuzzy multi-attribute group decision-making. Ye, J. (2015). Multiple attribute decision-making method based on the pos-
Journal of Computer and System Sciences, 80(1), 237256. sibility degree ranking method and ordered weighted aggregation oper-
Wan, S.P., Wang, F., Lin, L.L., & Dong, J.Y. (2015). An intuitionistic fuzzy lin- ators of interval neutrosophic numbers. Journal of Intelligent & Fuzzy
ear programming method for logistics outsourcing. Knowledge-Based Systems, 28(3), 13071317.
Systems, 82, 8094. Yue, Z.L. (2014). TOPSIS-based group decision-making methodology in
Wan, S.P., Xu, G.L., Wang, F., & Dong, J.Y. (2015). A new method for intuitionistic fuzzy setting. Information Sciences, 277, 141153.
Atanassovs interval-valued intuitionistic fuzzy MAGDM with incom- Zadeh, L.A. (1965). Fuzzy sets. Information and Control, 8(3), 338356.
plete attribute weight information. Information Sciences, 316, 329347. Zhang, H.M., & Yu, L.Y. (2012). MADM method based on cross-entropy
Wang, H.B., Smarandache, F., Zhang, Y.Q., & Sunderraman, R. (2005). Inter- and extended topsis with interval-valued intuitionistic fuzzy sets.
val neutrosophic sets and logic: Theory and applications in computing. Knowledge-Based Systems, 30, 115120.
Phoenix, AZ: Hexis. Zhang, H.Y., Wang, J.Q., & Chen, X.H. (2014). Interval neutrosophic
Wang, H.B., Smarandache, F., Zhang, Y.Q., & Sunderraman, R. (2010). Sin- sets and its application in multi-criteria decision-making prob-
gle valued neutrosophic sets. Multispace and Multistructure, 4, 410 lems. The Scientific World Journal. Advance online publication. doi:
413. 10.1155/2014/645953
Wang, J.Q., Han Z.Q., & Zhang, H.Y. (2014). Multi-criteria group decision- Zhang, H.Y., Ji, P., Wang, J.Q., & Chen, X.H. (2015a). An Improved weighted
making method based on intuitionistic interval fuzzy information. correlation coefficient based on integrated weight for interval neutro-
Group Decision and Negotiation, 23(4), 715733. sophic sets and its application in multi-criteria decision-making prob-
Wang, J.Q., & Li, X.E. (2015). The TODIM method with multi-valued neu- lems. International Journal of Computational Intelligence Systems, 8(6),
trosophic sets. Control and Decision, 30(6), 11391142 10271043.
Wang, J.Q., Peng, J.J., Zhang, H.Y., Liu, T., & Chen, X.H. (2015). An uncer- Zhang, H.Y., Wang, J.Q., & Chen, X.H. (2015b). An outranking approach
tain linguistic multi-criteria group decision-making method based on for multi-criteria decision-making problems with interval-valued neu-
a cloud model. Group Decision and Negotiation, 24(1), 171192. trosophic sets. Neutral Computing & Applications. Advance online pub-
Xu, Z.S. (2008). Dependent uncertain ordered weighted aggregation oper- lication. doi:10.1007/s00521-015-1882-3
ators. Information Fusion, 9(2), 310316. Zhao, X.Y., Tang, S., Yang, S.L., & Huang, K.D. (2013). Extended VIKOR
Xu, Z.S., & Da, Q.L. (2003). A possibility based method for priorities of method based on cross-entropy for interval-valued intuitionistic fuzzy
interval judgment matrices. Chinese Journal of Management Science, multiple criteria group decision-making. Journal of Intelligent & Fuzzy
11(1), 6365. Systems, 25(4), 10531066.