IOS Press
Abstract. Multi-valued neutrosophic sets (MVNSs) have recently become a subject of great interest for researchers, and have
been applied widely to multi-criteria decision-making (MCDM) problems. In this paper, the multi-valued neutrosophic geometric
weighted Choquet integral Heronian mean (MVNGWCIHM) operator, which is based on the Heronian mean and Choquet
integral, is proposed, and some special cases and the corresponding properties of the operator are discussed. Moreover, based on
the proposed operator, an MCDM approach for handling multi-valued neutrosophic information where the weights are unknown
is investigated. Furthermore, an illustrative example to demonstrate the applicability of the proposed decision-making approach
is provided, together with a sensitivity analysis and comparison analysis, which proves that its results are feasible and credible.
Keywords: Multi-criteria decision-making, Multi-valued neutrosophic sets, Choquet integral, Heronian mean
1876-1364/18/$17.00
c 2018 – IOS Press and the authors. All rights reserved
2 Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet
resolve MCDM problems. Ji et al. [22] developed a approach is established by combining the advantages
projection-based an acronym in Portuguese of inter- of the HM operator and Choquet integral to deal with
active and multi-criteria decision-making (TODIM) multi-valued neutrosophic MCDM problems where
method with multi-valued neutrosophic information. the weight information is completely unknown. An il-
Moreover, based on the definition of MVNSs, Peng et lustrative example is also provided to demonstrate the
al. [23] defined probability multi-valued neutrosophic applicability of the proposed method.
numbers (MVNNs), which are also an extension of The rest of paper is organized as follows. In Sec-
MVNSs. Wu and Wang [24] investigated some cross- tion 2, some basic concepts of NSs, MVNSs and the
entropy measures of MVNNs and applied them to operations of MVNNs are briefly reviewed. Then in
the problem of selecting a suitable middle-level man- Section 3, the multi-valued neutrosophic geometric
ager. Finally, Wang and Li [25] developed generalized weighted Choquet integral Heronian mean (MVNG-
single-valued neutrosophic hesitant fuzzy prioritized WCIHM) operator is proposed and some of the oper-
aggregation operators. ator’s special cases and corresponding properties are
The Choquet integral [26] and Heronian mean discussed. In Section 4, the way in which the extended
(HM) [27] are powerful tools for solving MCDM method can solve MCDM problems using MVNNs is
problems with correlated information in the decision- outlined. In Section 5, an illustrative example is pre-
making process. The Choquet integral can focus on sented to verify the proposed approach. Finally, con-
changing the weight vector of the aggregation oper- clusions are drawn in Section 6.
ator, while the HM can capture the interrelationships
of individual data. Recently, the two methods have
been applied widely to solve various MCDM prob- 2. Preliminaries
lems [28–40]. For example, Yager [28] introduced the
induced Choquet ordered averaging operator based on This section introduces the fuzzy measure and Cho-
the Choquet integral. Additionally, Yu et al. [30] de- quet integral, and reviews NSs and SNSs and MVNSs.
veloped hesitant Choquet integral fuzzy operators and Additionally, some operations of MVNNs, which will
applied them to MCDM problems. Wang et al. [35] be utilized in the later analysis, are also included.
proposed some Choquet integral operators with in-
terval 2-tuple linguistic information and also applied 2.1. The fuzzy measure and Choquet integral
them to MCDM problems. Liu and Zhang [38] defined
the neutrosophic hesitant fuzzy improved generalized Assume X = {x1 , x2 , . . . , xn } is the set of the cri-
weighted Heronian mean (NHFIGWHM) operator and teria and P (X) is the power set of X.
the neutrosophic hesitant fuzzy improved generalized Definition 1 [41, 42]. A fuzzy measure µ on the set X
geometric weighted Heronian mean (NHFIGGWHM) is a set function µ: P (X) → [0, 1] and satisfies the
operator. following axioms:
Based on the aforementioned studies, some attempts (1) µ (ϕ) = 0, µ (X) = 1;
have been made to define outranking relations, pref- (2) if B1 ⊆ B2 ⊆ X, then µ (B1 ) ≤ µ (B2 );
erence, aggregation operators and cross-entropy mea- (3) µ (B1 ∪ B2 ) = µ (B1 )+µ (B2 )+ρµ (B1 ) µ (B2 ),
sures of MVNNs. However, these methods cannot re- for ∀B1 , B2 ⊆ X, B1 ∩ B2 = ϕ, where ρ ∈
flect the interrelationships between the weights and in- (−1, +∞).
dividual data simultaneously. Moreover, for some ac- If ρ = 0, then (3) is reduced to the additive measure:
tual decision-making problems, each criterion has a re- ∀B1 , B2 ⊆ X, and B1 ∩ B2 = ϕ, µ (B1 ∪ B2 ) =
lationship with the other criteria. For example, if we µ (B1 ) + µ (B2 ).
want to select an investment project, we may con- If the elements of Bi are independent, then
sider the basic criteria to be risk and profit. As is well X
known, the higher the risk, the bigger the profit; there- µ (Bi ) = µ (xi ) for all Bi ⊆ X. (1)
fore, the two criteria are correlated with each other. xi ∈Bi
Clearly, the existing methods presented above can-
not resolve this type of decision-making problem. In If X is a finite set, then the ρ -fuzzy measure is rep-
order to address this shortcoming, we extended the resented as:
HM operator and Choquet integral to handle multi-
valued neutrosophic information. Consequently, a new
Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet 3
where (σ(1), σ(2), . . . , σ(n)) is a permutation of ∪ηψ ∈I˜ψ1 ,ηψ2 ∈I˜ψ2 {ηψ1 · ηψ2 } ,
1
MVNNs, which will be utilized in the latter analysis, ∪ξψ ∈F̃ψ1 ,ξψ2 ∈F̃ψ2 {ξψ1 + ξψ2 − ξψ1 · ξψ2 }i.
1
are introduced.
Definition 3 [15, 16]. Let X be a space of points (ob- Definition 5 [19]. Let ψ ∈ M V N N s, then the com-
jects), with a generic element in X denoted by x. An plement of an MVNN can be denoted by ψ C , which
MVNS ψ in X is characterized by can be defined as follows:
nD E o
x, T̃ψ (x) , I˜ψ (x) , F̃ψ (x) |x ∈ X , (4)
n o
ψ= ψ C = T̃ψC , I˜ψC , F̃ψC . (5)
where T̃ψ (x), I˜ψ (x), and F̃ψ (x) are three sets of pre- Here T̃ψC = ∪ξ∈F̃ψ {ξ}, I˜ψC = ∪η∈I˜ψ {1 − η} and
cise values in [0, 1] and are in the form of HFSs, de-
noting the truth-membership degree, indeterminacy- F̃ψC = ∪γ∈T̃ψ {γ}.
membership function and falsity-membership degree Definition 6 [19]. Let ψ1 and ψ2 be two MVNNs. The
respectively, and satisfying 0 ≤ γ, η, ξ ≤ 1, 0 ≤ comparision method can be defined as follows:
γ + +η + +ξ + ≤ 3, where γ ∈ T̃ψ (x) , η ∈ I˜ψ (x) , ξ ∈ (1) If s (ψ1 ) > s (ψ2 ) or s (ψ1 ) = s (ψ2 ) and
F̃ψ (x), γ + = sup T̃ψ (x), η + = sup I˜ψ (x) and a (ψ1 ) > a (ψ2 ), then ψ1 is superior to ψ2 , denoted by
ξ + = sup F̃ψ (x). ψ1 ψ2 ;
If X has only one element, then ψ is called an (2) If s (ψ1 ) = s (ψ2 ) and a (ψ1 ) = a (ψ2 ), then ψ1
MVNN, denoted by ψ = hT̃ψ (x) , I˜ψ (x) , F̃ψ (x)i. is indifferent to ψ2 , denoted by ψ1 ∼ ψ2 ;
For convenience, an MVNN can be denoted by ψ = (3) If s (ψ1 ) = s (ψ2 ) and a (ψ1 ) < a (ψ2 ) or
hT̃ψ , I˜ψ , F̃ψ i. The set of MVNNs are MVNNS. Ob- s (ψ1 ) < s (ψ2 ), then ψ1 is inferior to ψ2 , denoted by
viously, MVNSs are generally considered as an ex- ψ1 ≺ ψ2 ;
s (ψi ) = l ·l 1 ·l
P
tension of NSs. If each of T̃ψ (x) , I˜ψ (x) and F̃ψ (x) T̃ψ Ĩψ
γψ ∈T̃ψ ,ηψ ∈I˜ψ ,ξψ ∈F̃ψ
F̃ψ i i i i i i
i i i
for any x has only one value, i.e., γ, η and ξ, and (γψi −ηψi −ξψi ) 1
0 ≤ γ + η + ξ ≤ 3, then MVNSs are reduced to SNSs; 3 (i = 1, 2) and a (ψi ) = lT̃ ·lĨ ·lF̃
ψi ψi ψi
4 Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet
2
P (γψi +ηψi +ξψi ) m(m+1)
γψi ∈T̃ψi ,ηψi ∈I˜ψi ,ξψi ∈F̃ψi
3 (i = 1, 2) 1
m Y
Y m
represent the score function and accuracy function re- = (pZi + qZj ) . (7)
p + q i=1 j=i
spectively. Here γψi ∈ T̃ψi , ηψi ∈ I˜ψi and ξψi ∈ F̃ψi ;
lT̃ψ , lI˜ψ and lF̃ψ denote the number of elements in
i i i
Definition 9 Let ψi (i = 1, 2, . . . , m) be a group of
T̃ψ , I˜ψ and F̃ψ , respectively.
i i i
MVNNs, and µ be a fuzzy measure on X. Based on
the fuzzy measure, the MVNGWCIHM can be defined
3. The MVNGWCIHM operator as:
[
M V N GW CIHMp,q (ψ1 , ψ2 , . . . , ψm ) =
γψ ∈T̃ψ ,γ ∈T̃ψ
σ(i) σ(i) ψσ(j) σ(j)
ηψ ∈Ĩψ ,η ∈Ĩψ
σ(i) σ(i) ψσ(j) σ(j)
ξψ ∈F̃ψ ,ξψ ∈F̃ψ
σ(i) σ(i) σ(j) σ(j)
1
2
m(m+1) ! p+q
m Q
Q m pm(µ(Bσ(i) )−µ(Bσ(i+1) )) qm(µ(Bσ(j) )−µ(Bσ(j+1) ))
1− 1− 1 − 1 − γψσ(i) 1 − γψσ(j) ,
i=1 j=i
1
p+q
2
m m m(m+1)
pm(µ(Bσ(i) )−µ(Bσ(i+1) )) qm(µ(Bσ(j) )−µ(Bσ(j+1) ))
YY
1 − 1 − ηψσ(i) · ηψσ(j) ,
i=1 j=i
1
p+q
2
m Y m m(m+1)
pm(µ(Bσ(i) )−µ(Bσ(i+1) )) qm(µ(Bσ(j) )−µ(Bσ(j+1) ))
Y
1 − 1 − ξψσ(i) · ξψσ(j) . (9)
i=1 j=i
Here (σ(1), σ(2), . . . , σ(n)) is a permutation of Case 1 (1) If Eq. (1) holds, then µ({6(i)}) =
(1, 2, . . . , n), such that ψσ(1) ≤ ψσ(2) ≤ · · · ≤ ψσ(n) , µ(6(i)) − µ(6(i + 1)), i = 1, 2, . . . , m.
Bσ(j) = (σ(j), . . . σ(n)), and Bσ(n+1) = ∅.
Thus, Thus, Eq. (9) is reduced to a multi-valued neu-
The process of proof is omitted here.
Some special cases of the MVNGWCIHM operator trosophic weighted geometric Heronian mean (MVN-
are now discussed. WGHM) operator:
Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet 5
M V N W GHMp,q (ψ1 , ψ2 , . . . , ψm )
2 !
1 m m m(m+1)
= ⊗ ⊗ pmµ ({Xi }) ψσ(i) ⊕ qmµ ({Xj }) ψσ(j)
p+q i=1 j=i
1
2
! m(m+1) p+q
S Q m
m Q pmµ({Xi }) qmµ({Xj })
= 1 − 1 − 1 − 1 − γψσ(i) 1 − γψσ(j) ,
γψ ∈T̃ψ ,γ ∈T̃ψ
i=1 j=i
σ(i) ψσ(j)
σ(i) σ(j)
ηψ ∈Ĩψ ,η ∈Ĩψ
σ(i) σ(i) ψσ(j) σ(j)
ξψ ∈F̃ψ ,ξ ∈F̃ψ
σ(i) σ(i) ψσ(j) σ(j)
1
2
m(m+1) p+q
m m
YY
pmµ({Xi }) qmµ({X j })
1 − 1 − ηψσ(i) ηψσ(j) ,
i=1 j=1
1
2
m(m+1) p+q
m
YY m
pmµ({Xi }) qmµ({X j })
1 − 1 − ξψσ(i) ξψσ(j) . (10)
i=1 j=1
1
(2) If µ({Xi }) = m for all i = 1, 2, . . . , m, then Eq. (9) is reduced to a multi-valued neutrosophic averaging
geometric Heronian mean (MVNAGHM) operator:
2
m(m+1) !
1 m m 1 1
M V N AGHMp,q (ψ1 , ψ2 , . . . , ψm ) = ⊗ ⊗ pmψσ(i) ⊕ qmψσ(j)
p+q i=1 j=i m m
1
2
m(m+1) p+q
m Y m
[ Y p q
= 1 − 1 − 1 − 1 − γψσ(i) 1 − γψσ(j) , (11)
γψ ∈T̃ψ ,γψ ∈T̃ψ
i=1 j=i
σ(i) σ(i) σ(j) σ(j)
ηψ ∈Ĩψ ,η ∈Ĩψ
σ(i) σ(i) ψσ(j) σ(j)
ξψ ∈F̃ψ ,ξ ∈F̃ψ
σ(i) σ(i) ψσ(j) σ(j)
1 1
2
! m(m+1) p+q 2
! m(m+1) p+q
m Q
m
m Q
m
1 − ηψp σ(i) ηψq σ(j) 1 − ξψp σ(i) ξψq σ(j)
Q Q
1 − , 1 − .
i=1 j=1
i=1 j=1
P|B|
(3) If µ(B) = i=1 i for all B ⊆ X, (here |B| represents the number of the elements in the set B), then
i = µ(6(i)) − µ(6(i + 1)), i = 1, 2, . . . , m. Here ω = (ω1 , ω2 , . . . , ωm )T and i ≥ 0(i = 1, 2, . . . , m). In this case,
Eq. (9) is reduced to an MVNAGHM operator:
2 !
1 m m m(m+1)
M V N AGHMp,q (ψ1 , ψ2 , . . . , ψm ) = ⊗ ⊗ pmωi ψσ(i) ⊕ qmωj ψσ(j)
p+q i=1 j=i
1
m(m+1)2 p+q
m m
[ Y Y pmωi qmωj
= 1− 1− 1− 1 − γψσ(i) 1 − γψσ(j) ,
(12)
γψ ∈T̃ψ ,γψ ∈T̃ψ
i=1 j=i
σ(i) σ(i) σ(j) σ(j)
ηψ ∈Ĩψ ,η ∈Ĩψ
σ(i) σ(i) ψσ(j) σ(j)
ξψ ∈F̃ψ ,ξ ∈F̃ψ
σ(i) σ(i) ψσ(j) σ(j)
6 Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet
1 1
2
m(m+1) p+q 2
m(m+1) p+q
m m
m m
YY YY
pmω qmωj pmω qmωj
1 − 1−η i
η , 1− 1 − ξ i
ξ .
ψσ(i) ψσ(j)
ψσ(i) ψσ(j)
i=1 j=1
i=1 j=1
(4) In particular, if µ(B) = |B|m for all B ⊆ X, then both Eq. (9) and Eq. (11) are reduced to an MVNAGHM
operator.
Case 2 (1) If p = q = 21 , then the MVNGWCIHM operator, i.e. Eq. (9), is reduced to:
[
M V N GW CIHM 12 , 21 (ψ1 , ψ2 , . . . , ψm ) =
γψ ∈T̃ψ ,γ ∈T̃ψ
σ(i) σ(i) ψσ(j) σ(j)
ηψ ∈Ĩψ ,η ∈Ĩψ
σ(i) σ(i) ψσ(j) σ(j)
ξψ ∈F̃ψ ,ξ ∈F̃ψ
σ(i) σ(i) ψσ(j) σ(j)
(( 2
m(m+1) !)
m Q
m
q
Q m(µ(Bσ(i) )−µ(Bσ(i+1) )) m(µ(Bσ(j) )−µ(Bσ(j+1) ))
1− 1− 1− 1 − γψσ(i) 1 − γψσ(j) ,
i=1 j=i
2
! m(m+1)
m Y
m
r
m(µ(Bσ(i) )−µ(Bσ(i+1) )) m(µ(Bσ(j) )−µ(Bσ(j+1) ))
Y
1− 1− ηψσ(i) ηψσ(j) ,
i=1 j=i
2
! m(m+1)
m Y
m
r
m(µ(Bσ(i) )−µ(Bσ(i+1) )) m(µ(Bσ(j) )−µ(Bσ(j+1) ))
Y
1− 1 − ξψσ(i) ξψσ(j) . (13)
i=1 j=i
(2) If p = q = 1, then the MVNGWCIHM operator, i.e. Eq. (9), is reduced to:
[
M V N GW CIHM1,1 (ψ1 , ψ2 , . . . , ψm ) =
γψ ∈T̃ψ ,γ ∈T̃ψ
σ(i) σ(i) ψσ(j) σ(j)
ηψ ∈Ĩψ ,η ∈Ĩψ
σ(i) σ(i) ψσ(j) σ(j)
ξψ ∈F̃ψ ,ξ ∈F̃ψ
σ(i) σ(i) ψσ(j) σ(j)
v
u 2
! m(m+1)
m Q m
u Q m(µ(Bσ(i) )−µ(Bσ(i+1) )) m(µ(Bσ(j) )−µ(Bσ(j+1) ))
1 − t1 − 1 − 1 − γψσ(i) 1 − γψσ(j) ,
i=1 j=i
v
u 2
m(m+1)
u m Y m
m(µ(Bσ(i) )−µ(Bσ(i+1) )) m(µ(Bσ(j) )−µ(Bσ(j+1) ))
u Y
t1 − 1 − ηψσ(i) ηψσ(j) ,
i=1 j=1
v
u 2
m(m+1)
m Y m
u
u Y
m(µ(Bσ(i) )−µ(Bσ(i+1) )) m(µ(Bσ(j) )−µ(Bσ(j+1) ))
t1 − 1 − ξψσ(i) ξψσ(j) . (14)
i=1 j=1
(3) If q → 0, then the MVNGWCIHM operator, i.e. Eq. (9), is reduced to:
M V N GW CIHMp,0 (ψ1 , ψ2 , . . . , ψm )
2
! m(m+1) p1
m m+1−i
pm(µ(Bσ(i) )−µ(Bσ(i+1) ))
[ Y
= 1− 1−
1 − 1 − γψσ(i) ,
γψ ∈T̃ψ i=1
σ(i) σ(i)
ηψ ∈Ĩψ
σ(i) σ(i)
ξψ ∈F̃ψ
σ(i) σ(i)
Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet 7
2 p1
m m+1−i ! m(m+1)
pm(µ(Bσ(i) )−µ(Bσ(i+1) ))
Y
1 − 1 − ηψσ(i) ,
i=1
2 p1
m m+1−i ! m(m+1)
pm(µ(Bσ(i) )−µ(Bσ(i+1) ))
Y
1 − 1 − ξψσ(i) . (15)
i=1
(4) If p → 0, then the MVNGWCIHM operator, i.e. Eq. (9), is reduced to:
M V N GW CIHM0,q (ψ1 , ψ2 , . . . , ψm )
2
! m(m+1) q1
m i
qm(µ(Bσ(i) )−µ(Bσ(i+1) ))
[ Y
= 1 − 1 − 1 − 1 − γψσ(i) ,
γψ ∈T̃ψ i=1
σ(i) σ(i)
ηψ ∈Ĩψ
σ(i) σ(i)
ξψ ∈F̃ψ
σ(i) σ(i)
2 q1
m i ! m(m+1)
qm(µ(Bσ(i) )−µ(Bσ(i+1) ))
Y
1 − 1 − ηψσ(i) ,
i=1
2 q1
m i ! m(m+1)
qm(µ(Bσ(i) )−µ(Bσ(i+1) ))
Y
1 − 1 − ξψσ(i) . (16)
i=1
[
M V N W GCHMp,q (βi1 , βi2 , . . . , βim ) =
γσ(ii) ∈T̃σ(ii) ,γσ(ij) ∈T̃σ(ij)
ησ(ii) ∈Ĩσ(ii) ,ησ(ij) ∈Ĩσ(ij)
ξσ(ii) ∈F̃σ(ii) ,ξσ(ij) ∈F̃σ(ij)
1
2 p+q
m Q
m
! m(m+1)
Q pm(µ(Bσ(i) )−µ(Bσ(i+1) )) qm(µ(Bσ(j) )−µ(Bσ(j+1) ))
1 − 1 − 1 − 1 − γβσ(ii) 1 − γβσ(ij) ,
i=1 j=i
1
2
m(m+1) p+q
m m
pm(µ(Bσ(i) )−µ(Bσ(i+1) )) qm(µ(Bσ(j) )−µ(Bσ(j+1) ))
YY
1 − 1 − η η ,
βσ(ii) βσ(ij)
i=1 j=1
1
2
m(m+1) p+q
m m
YY
pm(µ(Bσ(i) )−µ(Bσ(i+1) )) qm(µ(Bσ(j) )−µ(Bσ(j+1) ))
1 − 1 − ξβσ(ii) ξβσ(ij) . (20)
i=1 j=1
According to the score function and accuracy func- In this section, an example is adapted from Wang
tion in Def. (3), we can obtain the final ranking and et al. [46] for further illustration. Hunan Nonferrous
Metals Holding Group Co. Ltd. is a large state-owned
select the best one(s). company whose main business is producing and sell-
Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet 9
ing nonferrous metals. It is also the largest manu- tors, namely: c1 : resources (such as the suitability of
facturer of multi-species nonferrous metals in China, the minerals and their exploration potential); c2 : poli-
with the exception of aluminum. In order to expand tics and policy (such as corruption and political risks);
its main business, the company is always engaged in c3 : economy (such as development vitality and stabil-
overseas investment, and a department which consists ity); and c4 : infrastructure (such as railway and high-
of executive managers and several experts in the field way facilities) are considered according to the depart-
has been established specifically to make decisions
ment’s previous investment projects. The evaluation of
on global mineral investment. Recently, the overseas
five candidates αi (i = 1, 2, 3, 4, 5) is performed us-
investment department has decided to select a pool
ing MVNNs by the three decision-makers under the
of alternatives from several foreign countries based
on preliminary surveys. Subsequently, the projects in criterion ck (k = 1, 2, 3, 4). One decision-maker could
those countries are to be investigated in detail. In this give several evaluation values for three membership
survey, the focus is on the first step of finding suit- degrees. In particular, in the case where two decision-
able candidate countries. Three countries (alternatives) makers set the same value, it is counted only once.
are taken into consideration, which are denoted by Then the multi-valued neutrosophic decision matrix
α1 , α2 , α3 , α4 , α5 . During the assessment, four fac- R = (αij )5×4 is constructed and shown as follows:
h{0.5, 0.7} , {0.3} , {0.3}i h{0.4} , {0.2, 0.3} , {0.1}i h{0.5} , {0.3, 0.4} , {0.4}i h{0.5, 0.6} , {0.2} , {0.3}i
h{0.3, 0.4} , {0.2} , {0.2}i h{0.6} , {0.2} , {0.3}i h{0.7} , {0.2} , {0.3}i h{0.4} , {0.2} , {0.1}i
R=
h{0.3} , {0.2, 0.3, 0.4} , {0.4}i h{0.4, 0.5} , {0.2.0.3} , {0.2}i h{0.6} , {0.1, 0.2} , {0.3}i h{0.5} , {0.1} , {0.2}i .
h{0.5} , {0.3} , {0.3}i h{0.6} , {0.1, 0.2} , {0.2}i h{0.7} , {0.2} , {0.2, 0.3}i h{0.6} , {0.2, 0.3, 0.4} , {0.2}i
h{0.6} , {0.2} , {0.4}i h{0.7} , {0.2, 0.3} , {0.4}i h{0.7} , {0.3} , {0.2}i h{0.5} , {0.1} , {0.3}i
5.1. An illustration of the proposed approach Which implies β51 < β52 < β54 < β53 such that
The procedures for obtaining the optimal alternative β5σ(1) = β51 , β5σ(2) = β52 ,
using the proposed method are shown in the following
β5σ(3) = β54 , β5σ(4) = β53 .
steps.
Step 1. Normalize the decision matrix.
Because all the criteria are of a maximizing type and Then
have the same measurement unit, there is no need for
ωσ(1) = µ Bσ(1) − µ Bσ(2)
normalization; thus, R̃ = (βij )4×4 = (αij )4×4 .
Step 2. Confirm the fuzzy measures on criteria set C. = µ (c1 , c2 , c3 , c4 ) − µ (c2 , c3 , c4 )
Assume that µ(c1 ) = 0.40, µ(c2 ) = 0.27, µ(c3 ) = = 1 − 0.78 = 0.22;
0.35 and µ(c4 ) = 0.30, then ρ = −0.58 can be
obtained. According to Eq. (1), µ(c1 , c2 ) = 0.55, ωσ(2) = µ (c2 , c4 , c3 ) − µ (c4 , c3 )
µ(c1 , c3 ) = 0.67, µ(c1 , c4 ) = 0.63, µ(c2 , c3 ) = 0.57, = 0.78 − 0.59 = 0.19;
µ(c2 , c4 ) = 0.52, µ(c3 , c4 ) = 0.59, µ(c1 , c2 , c3 ) =
0.83, µ(c1 , c2 , c4 ) = 0.80, µ(c2 , c3 , c4 ) = 0.78, ωσ(3) = µ (c4 , c3 ) − µ (c3 ) = 0.59 − 0.35 = 0.24;
µ(c1 , c3 , c4 ) = 0.85 and µ (c1 , c2 , c3 , c4 ) = 1 can be ωσ(4) = µ (c3 ) = 0.35.
determined.
Take β5j (j = 1, 2, 3, 4) for example, So ω5 = (0.22, 0.19, 0.35, 0.24).
Thus, the corresponding weight matrix can be ob-
s (β51 ) = 0, s (β52 ) = 0.0171, tained as:
Step 3. Aggregate all the performance values of each then the comprehensive values can be obtained as fol-
alternative. lows:
Utilizing the operator i.e., Eq. (20), to aggregate the
criterion values for each alternative and p = q = 2,
[
M V N GW CIHM2,2 (β11 , β12 , β13 , β14 ) =
γ1σ(i) ∈T̃1σ(i) ,γ1σ(j) ∈T̃1σ(j)
η1σ(i) ∈Ĩ1σ(i) ,η1σ(j) ∈Ĩ1σ(j)
ξ1σ(i) ∈F̃1σ(i) ,ξ1σ(j) ∈F̃1σ(j)
1
2 ! 2+2
4×5
4 Q
Q 4 2×4(µ(Bσ(i) )−µ(Bσ(i+1) )) 2×4(µ(Bσ(j) )−µ(Bσ(j+1) ))
1− 1− 1 − 1 − γβ1σ(i) 1 − γβ1σ(j) ,
i=1 j=i
1
2 2+2
4 Y 4 4×5
2×4(µ(Bσ(i) )−µ(Bσ(i+1) )) 2×4(µ(Bσ(j) )−µ(Bσ(j+1) ))
Y
1 − 1 − ηβ1σ(i) ηβ1σ(j) ,
i=1 j=i
1
2 2+2
4 4 4×5
2×4(µ(Bσ(i) )−µ(Bσ(i+1) )) 2×4(µ(Bσ(j) )−µ(Bσ(j+1) ))
YY
1 − 1 − ξβ1σ(i) ξβ1σ(j)
i=1 j=i
Then s(α5 ) > s(α1 ) > s(α4 ) > s(α2 ) > s(α3 )
can be obtained. Therefore, the final ranking is α5
M V N GW CIHM2,2 (β21 , β22 , β23 , β24 ) α1 α4 α2 α3 . The best alternative is α5 while
the worst is α3 .
= {{0.8311, 0.8494} , {0.3364} , {0.3214}} .
M V N GW CIHM2,2 (β31 , β32 , β33 , β34 ) 5.2. A sensitivity analysis
= {{0.8156, 0.8221}, {0.4074, 0.3983, 0.3929,
0.4004, 0.391, 0.3857, 0.3749, 0.3647,0.3587,
0.3671, 0.3566, 0.3506}, {{0.2935}}.
M V N GW CIHM2,2 (β41 , β42 , β43 , β44 )
= {{0.8819}, {0.3501, 0.3244, 0.3397, 0.3129,
0.3337, 0.3063}, {0.3244, 0.2941}}.
M V N GW CIHM2,2 (β51 , β52 , β53 , β54 )
= {{0.8835} , {0.3404, 0.3314} , {0.2869}} .
Fig. 1. The score values of alternatives and p = 0 and q ∈ (0, 10] .
Step 4. Rank all the alternatives.
The score values of each alternative(s) can be ob- In this subsection, the influence of different param-
tained as: eters on the ranking of alternatives is investigated by
using the MVNGWCIHM operator. Since the evalu-
s (α1 ) = 0.0817; s (α2 ) = 0.0608; ation values for three memberships in MVNNs are
sets of precise values in [0, 1], the different values of
s (α3 ) = 0.0488; s (α4 ) = 0.0816;
p, q ∈ (0, 10] are considered to conduct a sensitivity
s (α5 ) = 0.0869. analysis. The results are represented in Table 1 and Fig.
Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet 11
Table 1
The results by using the different parameters
Parameter (p, q) The final rankings The best alternative(s) The worst alternative(s)
p = q = 0.5 α5 α1 α4 α2 α3 α5 α3
p = 1, q = 0 α5 α4 α1 α3 α2 α5 α2
p = 0, q = 1 α1 α4 α5 α2 α3 α1 α3
p=q=1 α5 α1 α4 α2 α3 α5 α3
p=q=2 α5 α1 α4 α2 α3 α5 α3
p=q=5 α4 α1 α5 α2 α3 α4 α3
p = q = 10 α4 α1 α5 α2 α3 α4 α3
Table 2
The results obtained by utilizing the different methods
Methods The final ranking The best alternative(s) The worst alternative(s)
Ye [16] α5 α1 α4 α2 α3 α5 α3
Peng et al. [19] α5 α1 α4 α2 α3 α5 α3
Peng et al. [20] α1 , α5 α4 α2 α3 α1 , α 5 α3
Peng et al. [21] α1 α5 α4 α2 α3 α1 α3
Wang and Li [25] α5 α1 α4 α2 α3 α5 α3
Liu and Zhang [38] α5 α1 α4 α2 α3 α5 α3
The proposed method α5 α1 α4 α2 α3 α5 α3
values of p and q, the more emphasized the interac- Fig. 2. The score values of alternatives and p ∈ (0, 10] and q = 0.
more choices regarding the different values of the pa- 5.3. A comparison analysis
rameter, which are provided according to the decision-
In order to validate the feasibility of the proposed
makers’ preferences. decision-making method, a comparative study was
12 Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet
conducted with other methods; specifically those in relationships between the criteria and data should be
Ye [16], Peng et al. [19–21], Wang and Li [25] and Liu considered. Therefore, the main advantages of the pro-
and Zhang [38]. posed approach are not only its ability to deal effec-
To facilitate a comparison analysis, the same exam- tively with the preference information expressed by
ple that was used in Section 5 is used here as well. MVNNs, but also its consideration that the weights of
Since the compared methods presented above cannot the criteria and individual data are interrelated, which
handle multi-valued neutrosophic information where makes the final results correspond better with actual
the weight is completely unknown, we now use the decision-making problems.
same example but with the weights of criteria deter-
mined as w = (0.25, 0.21, 0.35, 0.19). Subsequently,
the proposed method is reduced to the methods used in 6. Conclusions
Liu and Zhang [38]. For the proposed method, we can
use the determined weight directly in Step 3 in Sub- For real decision-making problems, the criteria
section 5.1 and p = q = 2. Then for the method in and an individual’s evaluation are always interre-
Ye [16] and Peng et al. [19], the weighted averaging lated. MVNNs can be used widely to deal effec-
operator and the weighted arithmetic power averaging tively with uncertain, imprecise and inconsistent in-
operator are utilized respectively. For the method in formation. Based on the HM and Choquet integral,
Wang and Li [25], we assume that the criterion satisfy the MVNGWCIHM operator has been proposed in
C1 C2 C3 C4 and λ = 1. Moreover, the gener- this paper, and some special cases and the corre-
alized single-valued neutrosophic hesitant fuzzy prior- sponding properties of the operator were discussed.
itized weighted geometric operator is used to deal with Moreover, based on the MVNGWCIHM operator, a
the same example. Consequently, the final results can multi-valued neutrosophic approach was investigated
be obtained as shown in Table 2. to resolve MCDM problems where the data is in the
Based on the results presented in Table 4, we can see form of MVNNs and the weights of criteria are com-
that the results from the proposed approach are consis- pletely unknown. Additionally, an illustrative example
tent with those that use the methods in Ye [16], Peng demonstrated the application of the proposed decision-
et al. [19], Wang and Li [25] and Liu and Zhang [38]; making approach, and proved that its results are fea-
the best alternative is α5 while the worst is α3 . For the sible and credible. The main advantages of the pro-
other compared methods presented in Refs. [20, 21], posed approach are its ability to consider effectively
although there is a slight difference in the final rank- the interrelationships of the criteria and data and that
ings of these methods, the alternative α1 is always the the weights of the criteria are completely unknown,
best one(s). which makes the final results better correspond with
Based on the results presented above, some con- actual decision-making problems. In future research,
clusions can be drawn and these are now discussed. the measures of MVNNs will be further investigated.
Firstly, although the result of the proposed approach is
consistent with that using the method of Ye [16], Peng
et al. [19], Wang and Li [25] and Liu and Zhang [38], Acknowledgement
these four methods are unable to consider the data in-
terrelationships of the criterion. Secondly, the com- This work is supported by the National Natural
pared methods mentioned above cannot resolve multi- Science Foundation of China (Nos. 71701065 and
valued neutrosophic problems where the weight infor- 71571193), the Postdoctoral Science Foundation of
mation is completely unknown. Thirdly, the methods China (No. 2017M610511), the Humanities and So-
developed by Peng et al. [20] can only resolve MCDM cial Sciences Foundation of Ministry of Education of
problems in which the number of criteria clearly ex- China (No. 15YJCZH127), and the Natural Science
ceeds the number of alternatives; while the method Foundation of Hubei Province (No. 2017CFC852).
in Peng et al. [21] is better used in resolving prob-
lems with a large number of alternatives and few cri-
teria. Otherwise, the final results cannot be obtained Conflict of interest
directly. However, the approach proposed in this pa-
per is the optimal method for MCDM problems where The authors declare that there is no conflict of inter-
the weight of criteria is completely unknown, and the est regarding the publication of this paper.
Juan-juan Peng et al. / Multi-criteria Decision-making Approach based on Multi-valued Neutrosophic Geometric Weighted Choquet 13