0 valutazioniIl 0% ha trovato utile questo documento (0 voti)

0 visualizzazioni10 pagineDec 30, 2019

© © All Rights Reserved

PDF, TXT o leggi online da Scribd

© All Rights Reserved

0 valutazioniIl 0% ha trovato utile questo documento (0 voti)

0 visualizzazioni10 pagine© All Rights Reserved

Sei sulla pagina 1di 10

DOI:10.3233/IFS-162179

IOS Press

attribute decision making under simplified

neutrosophic environments

Jun Ye∗

Department of Electrical and Information Engineering, Shaoxing University, Shaoxing,

Zhejiang Province, P.R. China

Abstract. A simplified neutrosophic set (SNS) is a subclass of neutrosophic set and contains a single-valued neutrosophic

set (SVNS) and an interval neutrosophic set (INS). It was proposed as a generalization of an intuitionistic fuzzy set (IFS)

and an interval-valued intuitionistic fuzzy set (IVIFS) in order to deal with indeterminate and inconsistent information. The

paper proposes another form of the Dice measures of SNSs and the generalized Dice measures of SNSs and indicates that

the Dice measures and asymmetric measures (projection measures) are the special cases of the generalized Dice measures in

some parameter values. Then, we develop the generalized Dice measures-based multiple attribute decision-making methods

with simplified neutrosophic information. By the weighted generalized Dice measures between each alternative and the ideal

solution (ideal alternative) corresponding to some parameter value required by decision makers’ preference, all the alternatives

can be ranked and the best one can be obtained as well. Finally, a real example on the selection of manufacturing schemes

demonstrates the applications of the proposed decision-making methods under simplified neutrosophic environment. The

effectiveness and flexibility of the proposed decision-making methods are shown by choosing different parameter values.

Keywords: Generalized Dice measure, Dice measure, decision making, simplified neutrosophic set, asymmetric measure,

projection measure

problems with indeterminate and inconsistent infor-

Multiple attribute decision making is a main branch mation, which IFSs and IVIFSs cannot describe and

of decision theory, where neutrosophic theory intro- deal with. Recently, many researchers have applied

duced by Smarandache [1] has been successfully SNSs and the subclasses of SNSs (SVNSs and INSs)

applied in recent years. As a generalization of an to the decision-making problems. Various methods

intuitionistic fuzzy set (IFS) [16] and an interval- have been developed to solve the multiple attribute

valued intuitionistic fuzzy (IVIFS) [17], a simplified decision-making problems with simplified neutro-

neutrosophic set (SNS) introduced by Ye [10] is sophic information. For example, Ye [9] proposed

a subclass of a neutrosophic sets [1], including a the correlation coefficient of SVNSs and applied it

single-valued neutrosophic set (SVNSs) [4] and an to multiple attribute decision making. Chi and Liu

interval neutrosophic set (INSs) [3]. Hence, SNSs [18] and Biswas et al. [21] extended TOPSIS method

to single-valued and interval neutrosophic multiple

∗ Corresponding author. Jun Ye, Department of Electrical and

attribute decision-making problems. Ye [11–13] pre-

Information Engineering, Shaoxing University, 508 Huancheng

West Road, Shaoxing, Zhejiang Province 312000, P.R. China. Tel.:

sented some similarity measures of SVNSs, INSs

+86 575 88327323; E-mail: yehjun@aliyun.com. and SNSs and applied them to decision making.

1064-1246/16/$35.00 © 2016 – IOS Press and the authors. All rights reserved

664 J. Ye / The generalized Dice measures for multiple attribute decision making

Ye [14] put forward a cross-entropy measure of is given to show the application of the proposed meth-

SVNSs for multiple attribute decision making prob- ods, and then the effectiveness and flexibility of the

lems. Ye [10], Zhang et al. [5], Liu et al. [19], Liu and proposed methods are indicated by choosing different

Wang [20], and Peng et al. [8] developed some simpli- parameter values. Finally, Section 7 contains conclu-

fied, interval and single-valued neutrosophic number sions and future work.

aggregation operators and applied them to multiple

attribute decision-making problems. Peng et al. [7]

and Zhang et al. [6] proposed outranking approaches 2. The Dice measures of SNSs

for multicriteria decision-making problems with sim-

plified and interval neutrosophic information. Sahin As a subset of a neutrosophic set [1], Ye [10] intro-

and Kucuk [22] presented a subsethood measure for duced a SNS and gave its definition.

SVNSs and applied it to multiple attribute decision

making. Şahin and Liu [23] introduced a maxi- Deﬁnition 1. [10] A SNS S in the universe of

mizing deviation method for neutrosophic multiple discourse X is defined as S = {x, tS (x), uS (x), v S (x)

attribute decision making with incomplete weight |x ∈ X}, where tS (x): X → [0, 1], us (x) : X →

information. Ye [15] presented a multiple attribute [0, 1], and v s (x) : X → [0, 1] are a truth-membership

decision-making method based on the possibility function and an indeterminacy-membership func-

degree ranking method and ordered weighted aggre- tion, a falsity- membership function, respectively,

gation operators of interval neutrosophic numbers. of the element x to the set S with the condition

Since the Dice measure is one of vector similarity 0 ≤ ts (x) + us (x) + v s (x) ≤ 3 for x ∈ X.

measures, it is a useful mathematical tool for han- In fact, SNSs contain the concepts of SVNSs

dling decision-making problems. However, the Dice and INSs, which are the subclasses of SNSs. For

measure of SNSs [13] used for decision making lacks convenience, a component element < x, ts (x),

flexibility in decision-making process. Therefore, it us (x), v s (x) > in a SNS S is denoted by sx =<

is necessary to improve the Dice measure of SNSs tx , ux , v x > for short, which is called the simpli-

to handle multiple attribute decision-making prob- fied neutrosophic number (SNN), where tx , ux , v x ∈

lems to satisfy the requirements of decision makers’ [0, 1] and 0 ≤ tx , +ux + v x ≤ 3 for a single-

preference and flexible decision making. In order valued neutrosophic number (SVNN), and then

to do so, the main purposes of this paper are: (1) tx = [txL , txU ] ⊆ [0, 1], ux = [uL U

x , ux ] ⊆ [0, 1], v x =

to propose another form of the Dice measures of [v x , v x ] ⊆ [0, 1] and 0 ≤ tx + ux + v U

L U U U

x ≤ 3 for an

SNSs, (2) to present the generalized Dice measures interval neutrosophic number (INN).

of SNSs, and (3) to develop the generalized Dice Ye [13] presented the Dice measures of SNSs,

measures-based multiple attribute decision-making which was defined below.

methods with simplified neutrosophic information.

In the decision making process, the main advan- Deﬁnition 2. [13] Let S1 = {s11 , s12 , . . . , s1n } and

tage of the proposed methods is more general and S2 = {s21 , s22 , . . . , s2n } be two SNSs. If s1j =<

more flexible than existing decision-making methods t1j , u1j , v 1j , > and s2j =< t2j , u2j , v 2j , > (j =

with simplified neutrosophic information to satisfy 1, 2, . . . , n) are the j-th SVNNs in S1 and S2 respec-

the decision makers’ preference and/or practical tively, then the Dice measure between S1 and S2 is

requirements. defined as:

The rest of the paper is organized as follows.

Section 2 reviews the Dice measures of SNSs. Sec- n

1 2s1j · s2j

tion 3 proposes another form of the Dice measures DSVNN1 (S1 , S2 ) = 2 2 .

of SNSs. In Section 4, we propose the general- n s1j + s2j

j=1

ized Dice measures of SNSs and indicate the Dice n

1 2(t1j t2j + u1j u2j + v 1j v 2j )

measures and asymmetric measures (projection mea- =

sures) as the special cases of the generalized Dice n (t 2 + u2 + v 2 ) + (t 2 + u2 + v 2 )

j=1 1j 1j 1j 2j 2j 2j

measures in some parameter values. In Section 5, the (1)

generalized Dice measures-based multiple attribute If s1j =< t1j , u1j , v 1j > and s2j =< t2j , u2j ,

decision-making methods are developed under sim- v 2j > (j = 1, 2, . . . , n) are the j-th INNs in S1 and

plified neutrosophic environment. In Section 6, a real S2 respectively, then the Dice measure between S1

example on the selection of manufacturing schemes and S2 is defined as:

J. Ye / The generalized Dice measures for multiple attribute decision making 665

1 2s1j · s2j

n 3. Another form of the Dice measures of SNSs

DINN1 (S1 , S2 ) = 2 2 .

n s1j + s2j

j=1 This section proposes another form of the Dice

L L U U measures of SNSs, which is defined as follows.

t1j t2j + t1j t2j + uL1j uL2j

2 Deﬁnition 3. Let S1 = {s11 , s12 , . . . , s1n } and

1

n

+uU1j uU2j + v L1j v L2j + v U1j v U2j

= ⎛ ⎞ (2) S2 = {s21 , s22 , . . . , s2n } be two SNSs. If s1j =<

L 2

n

j=1

(t1j ) + (uL1j )2 + (v L1j )2 t1j , u1j , v 1j > and s2j =< t2j , u2j , v 2j > (j =

⎜ ⎟ 1, 2, . . . , n) are the j-th SVNNs in S1 and S2 respec-

⎜ +(t1j ) + (uU1j )2 + (v U1j )2 ⎟

U 2

⎜ ⎟ tively, then the Dice measure between S1 and S2 is

⎜ ⎟

⎜ +(t L )2 + (uL )2 + (v L )2 ⎟ defined as:

⎝ 2j 2j 2j ⎠

U 2

+(t2j ) + (uU2j )2 + (v U2j )2

2(S1 · S2 )

DSVNN2 (S1 , S2 ) =

Then, the two Dice measures DSVNN1 (S1 , S2 ) and |S1 |2 + |S2 |2

n

DINN1 (S1 , S2 ) satisfy the following properties [13]: 2 j=1 (t1j t2j + u1j u2j + v 1j v 2j )

(P1) DSVNN1 (S1 , S2 ) = DSVNN1 (S2 , S1 ) and = n n .

j=1 (t1j + u1j + v 1j ) + j=1 (t2j + u2j + v 2j )

2 2 2 2 2 2

DINN1 (S1 , S2 ) = DINN1 (S2 , S1 );

(5)

(P2) 0 ≤ DSVNN1 (S1 , S2 ) ≤ 1 and 0 ≤ DINN1

(S1 , S2 ) ≤ 1;

If s1j =< t1j , u1j , v 1j > and s2j =< t2j , u2j ,

(P3) DSVNN1 (S1 , S2 ) = 1 and DINN1 (S1 , S2 ) =

v 2j > (j = 1, 2, . . . , n) are the j-th INNs in S1 and

1, if S1 = S2 .

S2 respectively, then the Dice measure between S1

Especially when tij = tijL = tijU , uij = uL U

ij = uij , and S2 is defined as:

U

and v ij = v Lij = v ij for i = 1, 2 and j = 1, 2, . . . , n

are hold, Equation (2) is degenerated to Equation (1).

In real applications, one usually takes the 2(S1 · S2 )

DINN2 (S1 , S2 ) =

important differences of each element sij (i = |S1 |2 + |S2 |2

1, 2; j = 1, 2, . . . , n) into account. Let W = ⎛ ⎞

L L U U

n t1j t2j + t1j t2j + uL L

1j u2j

(w1 , w2 , . . . , wn )T be the weight vector

for sij (i = 2 ⎝ ⎠

j=1

1, 2; j = 1, 2, . . . , n), wj ≥ 0 and nj=1 wj = 1. +uU u U

+ v L L

v + v U U

v

1j 2j 1j 2j 1j 2j

Then, based on Equations (1) and (2), Ye [13] fur- =⎛ ⎡ ⎤ ⎞.

L 2 U 2 L 2

ther introduced the weighted Dice measures of SNSs, n (t1j ) + (t1j ) + (u1j )

⎜ ⎣ ⎦+⎟

respectively, as follows: ⎜ j=1 ⎟

⎜ +(uU )2 + (v L )2 + (v U )2 ⎟

n ⎜ 1j 1j 1j ⎟

2s1j · s2j ⎜ ⎡ ⎤ ⎟

DWSVNN1 (S1 , S2 ) = w j 2 2 , ⎜ L 2 U 2

(t2j ) + (t2j ) + (u2j )L 2 ⎟

⎜ n ⎟

s1j + s2j ⎝ ⎣ ⎦ ⎠

j=1

j=1 U 2 L 2 U 2

n

+(u2j ) + (v 2j ) + (v 2j )

2(t1j t2j + u1j u2j + v 1j v 2j )

= wj (6)

j=1

2

(t1j + u21j + v 21j ) + (t2j

2 + u2 + v 2 )

2j 2j Obviously, the two Dice measures DSVNN2

(3) (S1 , S2 ) and DINN2 (S1 , S2 ) also satisfy the follow-

ing properties:

n

2s1j · s2j (P1) DSVNN2 (S1 , S2 ) = DSVNN2 (S2 , S1 ) and

DWINN1 (S1 , S2 ) = wj 2 2

s1j + s2j DINN2 (S1 , S2 ) = DINN2 (S2 , S1 );

j=1

(P2) 0 ≤ DSVNN2 (S1 , S2 ) ≤ 1 and 0 ≤ DINN2

L L U U

t1j t2j + t1j t2j + uL1j uL2j (S1 , S2 ) ≤ 1;

2 (P3) DSVNN2 (S1 , S2 ) = 1 and DINN2 (S1 , S2 ) =

n

+uU1j uU2j + v L1j v L2j + v U1j v U2j

= wj ⎛ ⎞ . (4) 1, if S1 = S2 .

L 2

j=1

(t1j ) + (uL1j )2 + (v L1j )2

⎜ ⎟ Proof:

⎜ +(t1jU 2

) + (uU1j )2 + (v U1j )2 ⎟

⎜ ⎟ (P1) It is obvious that the property is true.

⎜ ⎟

⎜ +(t L )2 + (uL )2 + (v L )2 ⎟ (P2) It is obvious that the property is true accord-

⎝ 2j 2j 2j ⎠ ing to the inequality a2 + b2 ≥ 2ab for Equations (5)

U 2

+(t2j ) + (uU2j )2 + (v U2j )2 and (6).

666 J. Ye / The generalized Dice measures for multiple attribute decision making

n

(t1j t2j + u1j u2j + v 1j v 2j )

n) and |S1 | = |S2 |. So there are DSVNN2 (S1 , S2 ) = 1 = ,

n λ(t1j

2

+ u21j + v 21j ) + (1 − λ)(t2j

2

+ u22j + v 22j )

and DINN2 (S1 , S2 ) = 1. j=1

(i = 1, 2; j = 1, 2, . . . , n) have different weights.

Let W = (w1 , w2 , . . . , wn )T be the weight vec- S1 · S2

nfor sij (i = 1, 2; j = 1, 2, . . . , n), wj ≥ 0 and

tor GSVNN2 (S1 , S2 ) =

λ |S1 | + (1 − λ) |S2 |2

2

j=1 wj = 1. Then, based on Equations (5) and (6) n

we further introduce the weighted Dice measures of (t t + u1j u2j

j=1 1j 2j

+ v 1j v 2j )

= n n ,

SNSs, respectively, as follows: λ (t 2 + u1j + v 21j ) + (1 − λ)

2 2

(t2j + u22j + v 22j )

j=1 1j j=1

(10)

2(S1 · S2 )w where λ is a positive parameter for 0 ≤ λ ≤ 1.

DWSVNN2 (S1 , S2 ) =

|S1 |2w + |S2 |2w Then, the generalized Dice measures imply some

n special cases by choosing some values of the param-

2 j=1

w2j (t1j t2j + u1j u2j + v 1j v 2j )

= n n , eter λ. If λ = 0.5, the two generalized Dice measures

w2 (t 2

j=1 j 1j

+ u21j + v 21j ) + j=1

w2j (t2j

2

+ u22j + v 22j )

(7)

(9) and (10) are degenerated to the Dice measures (1)

and (5); if λ = 0, 1, the two generalized Dice mea-

sures are degenerated to the following asymmetric

2(S1 · S2 )w measures respectively:

DWINN2 (S1 , S2 ) =

|S1 |2w + |S2 |2w n

⎛ ⎞ 1 s1j · s2j

L L U U GSVNN1 (S1 , S2 ) = 2

n t1j t2j + t1j t2j + uL L

1j u2j n s2j

2⎝ ⎠ j=1

2 j=1 wj

+uU U

1j u2j + vL L

1j v 2j + vU U

1j v 2j 1

n

t1j t2j + u1j u2j + v 1j v 2j

=⎛ ⎡ ⎤ ⎞. = for λ = 0, (11)

L 2 U 2

n (t1j ) + (t1j ) + (uL

1j )

2

n t2j

2 + u2 + v 2

j=1

⎜ w2 ⎣ ⎦+⎟ 2j 2j

⎜ j=1 j ⎟

⎜ +(uU )2

+ (v L 2

) + (v U 2

) ⎟ n

1 s1j · s2j

⎜ 1j 1j 1j ⎟

⎜ ⎡ ⎤ ⎟ GSVNN1 (S1 , S2 ) = 2

⎜ L 2 U 2

(t2j ) + (t2j ) + (u2j )L 2 ⎟ n s1j

⎜ n ⎟ j=1

⎝ w2j ⎣ ⎦ ⎠

n

j=1

+(uU L 2 U 2

2j ) + (v 2j ) + (v 2j )

2

1 t1j t2j + u1j u2j + v 1j v 2j

= for λ = 1, (12)

(8) n t1j

2 + u2 + v 2

j=1 1j 1j

S1 · S2

GSVNN2 (S1 , S2 ) =

4. The generalized Dice measures of SNSs |S2 |2

n

j=1 (t1j t2j +u1j u2j + v 1j v 2j )

In this section, we propose the generalized Dice = n for λ = 0, (13)

j=1 (t2j + u2j + v 2j )

2 2 2

measures of SNSs to extend the Dice measures of

SNSs. S1 · S2

As the generalization of the Dice measures of GSVNN2 (S1 , S2 ) =

|S1 |2

SNSs, the generalized Dice measures between SNSs n

are defined below. j=1 (t1j t2j +u1j u2j + v 1j v 2j )

= n for λ = 1. (14)

j=1 (t1j + u1j + v 1j )

2 2 2

Deﬁnition 4. Let S1 = {s11 , s12 , . . . , s1n } and

S2 = {s21 , s22 , . . . , s2n } be two SNSs, where Obviously, the four asymmetric measures are the

s1j = (t1j , u1j , v 1j ) and s2j = (t2j , u2j , v 2j ) (j = extension of the relative projection measure (the

1, 2, . . . , n) are considered as the j-th SVNNs in improved projection measure) of interval numbers

the SNSs S1 and S2 . Then the generalized Dice mea- [2], hence the four asymmetric measures can be con-

sures between S1 and S2 are defined, respectively, as

sidered as the projection measures of SNSs.

follows:

For practical applications, the elements of sij

(i = 1, 2; j = 1, 2, . . . , n) imply different weights.

1

n

s1j · s2j

GSVNN1 (S1 , S2 ) = 2 2 Assume that W = (w1 , w2 . . . , wn )T is the weight

n λ s1j + (1 − λ) s2j

j=1 vector for sij (i = 1, 2; j = 1, 2, . . . , n), wj ≥ 0

J. Ye / The generalized Dice measures for multiple attribute decision making 667

and nj=1 wj = 1. Thus, based on Equations (9) and where λ is a positive parameter for 0 ≤ λ ≤ 1. Espe-

(10) we further introduce the following weighted cially, when tij = tijL = tijU , uij = uL U

ij = uij , and v ij =

U

generalized Dice measures of SNSs, respectively, as vL

ij = v ij for i = 1, 2 and j = 1, 2, . . . , n are hold,

follows: Equations (17) and (18) are degenerated to Equations

n

s1j · s2j

(9) and (10).

GWSVNN1 (S1 , S2 ) = wj 2 2 Similarly, if λ = 0.5, the two generalized Dice

j=1

λ s1j + (1 − λ) s2j

measures (17) and (18) are degenerated to the Dice

n measures (2) and (6); if λ = 0, 1, then the two gener-

t1j t2j + u1j u2j + v 1j v 2j

= wj , alized Dice measures are degenerated to the following

λ(t1j

2

+ u21j + v 21j ) + (1 − λ)(t2j

2

+ u22j + v 22j )

j=1 asymmetric measures respectively:

(15)

n n

1 s1j · s2j 1

GINN3 (S1 , S2 ) = 2 =

n j=1 s2j n j=1

(S1 · S2 )w

GWSVNN2 (S1 , S2 ) = U U

λ |S1 |2w + (1 − λ) |S2 |2w

L L

t1j t2j + t1j t2j + uL1j uL2j + uU1j uU2j + v L1j v L2j + v U1j v U2j

n L 2

(t2j U 2

) + (uL2j )2 + (v L2j )2 + (t2j ) + (uU2j )2 + (v U2j )2

j=1 wj (t1j t2j + u1j u2j + v 1j v 2j )

2

= ⎛ n ⎞. (16) for λ = 0, (19)

λ w2j (t1j

2

+ u21j + v 21j )+

⎜ j=1 ⎟ n

s1j · s2j

n

⎝ n ⎠ 1 1

GINN3 (S1 , S2 ) = 2 =

(1 − λ) w2 (t 2 + u22j + v 22j ) n s1j n

j=1 j 2j j=1 j=1

U U

L L

t1j t2j + t1j t2j + uL1j uL2j + uU1j uU2j + v L1j v L2j + v U1j v U2j

Deﬁnition 5. Let S1 = {s11 , s12 , . . . , s1n } and U 2

S2 = {s21 , s22 , . . . , s2n } be two SNSs, where

L 2

(t1j ) + (uL1j )2 + (v L1j )2 + (t1j ) + (uU1j )2 + (v U1j )2

s1j = (t1j , u1j , v 1j ) and s2j = (t2j , u2j , v 2j ) (j = for λ = 1, (20)

1, 2, . . . , n) are considered as the j-th INNs in the

SNSs S1 and S2 . Then the generalized Dice mea-

sures between S1 and S2 are defined, respectively,

S1 · S2

as follows: GINN4 (S1 , S2 ) =

|S2 |2

1 n L L U U

n

s1j · s2j 1 (t t + t1j t2j + uL1j uL2j + uU1j uU2j + v L1j v L2j + v U1j v U2j )

GINN3 (S1 , S2 ) = 2 2 = j=1 1j 2j

n λ s1j + (1 − λ) s2j n = n L 2 U 2 L 2 U 2 L 2 U 2

j=1

j=1

[(t2j ) + (t2j ) + (u2j ) + (u2j ) + (v 2j ) + (v 2j ) ]

n L L

t1j t2j + t U t2j

U

+ uL1j uL2j + uU1j uU2j + v L1j v L2j + v U1j v U2j for λ = 0, (21)

⎛

1j

L 2

⎞ ,

(t1j ) + (uL1j )2 + (v L1j )2 S1 · S2

j=1 GINN4 (S1 , S2 ) =

⎜λ ⎟ |S1 |2

⎜ U 2

+(t1j ) + (uU1j )2 + (v U1j )2 ⎟ n L L U U

⎜ ⎟ (t t + t1j t2j + uL1j uL2j + uU1j uU2j + v L1j v L2j + v U1j v U2j )

⎜ L 2 ⎟

⎜ L 2 L 2

(t2j ) + (u2j ) + (v 2j ) ⎟ j=1 1j 2j

= n

⎝ ⎠ L 2 U 2 L 2 U 2 L 2 U 2

[(t1j ) + (t1j ) + (u1j ) + (u1j ) + (v 1j ) + (v 1j ) ]

+(1 − λ) j=1

U 2

+(t2j ) + (uU2j )2 + (v U2j )2 for λ = 1. (22)

(17)

Then, the four asymmetric measures are also con-

S1 · S2

GINN4 (S1 , S2 ) = sidered as the extension of the relative projection

λ |S1 |2 + (1 − λ) |S2 |2

n measure (the improved projection measure) of inter-

U U

j=1

L L

(t1j t2j + t1j t2j + uL1j uL2j + uU1j uU2j + v L1j v L2j + v U1j v U2j ) val numbers [2], which are also called the projection

= ⎛ ⎞ ,

measures of SNSs.

n L 2

(t1j U 2

) + (t1j ) + (uL1j )2

⎜λ + ⎟ For practical applications, the elements of sij (i =

⎜ j=1 +(uU1j )2 + (v L1j )2 + (v U1j )2 ⎟ 1, 2; j = 1, 2, . . . , n) imply different weights.

⎜ ⎟

⎜ ⎟ Assume that W = (w1 , w2 , . . . , wn )T is the weight

⎜ n L 2

(t2j U 2

) + (t2j ) + (uL2j )2 ⎟

⎝ ⎠ vector sij (i = 1, 2; j = 1, 2, . . . , n), wj ≥ 0

(1 − λ) for

n

+(uU2j )2 + (v L2j )2 + (v U2j )2 j=1 wj = 1. Similarly, based on Equations

j=1 and

(18) (17) and (18) we also further introduce the weighted

668 J. Ye / The generalized Dice measures for multiple attribute decision making

generalized Dice measures of SNSs, respectively, as be used to help identify the best alternative in the

follows: decision set [13]. Hence, by an ideal SVNN

n

s1j · s2j

GWINN3 (S1 , S2 ) = wj 2 2 sj∗ =< tj∗ , u∗j , v ∗j >=< max(tij ), min(uij ), min(v ij ) >

j=1 λ s1j + (1 − λ) s2j i i i

L L U U

t1j t2j + t1j t2j + uL1j uL2j or an ideal INN

n

+uU1j uU2j + v L1j v L2j + v U1j v U2j sj∗ =< tj∗ , u∗j , v ∗j >=< [max(tijL ), max(tijU )],

= wj ⎛ L 2

⎞, i i

j=1

(t1j ) + (uL1j )2 + (v L1j )2

⎜λ ⎟ [min(uL U L U

ij ), min(uij )], [min(v ij ), min(v ij )] >

⎜ U 2

+(t1j ) + (uU1j )2 + (v U1j )2 ⎟ i i i i

⎜ ⎟

⎜ L 2 ⎟

⎜ (t ) + (uL 2

) + (v L 2

) ⎟ for j = 1, 2, . . . , n and i = 1, 2, . . . , m, we can

⎝ 2j 2j 2j ⎠

+(1 − λ) determine a simplified neutrosophic ideal solu-

U 2

+(t2j ) + (uU2j )2 + (v U2j )2 tion (ideal alternative) S ∗ = {s1∗ , s2∗ , . . . , sn∗ }, where

(23)

sj∗ =< tj∗ , u∗j , v ∗j > is the j-th ideal SNN.

In the decision-making process, decision makers

(S1 · S2 )w take some value of the parameter λ ∈ [0, 1] accord-

GWINN4 (S1 , S2 ) =

λ |S1 |2w + (1 − λ) |S2 |2w ing to their preference and/or real requirements, the

L L U U

weighted generalized Dice measure between Si (i =

n t1j t2j + t1j t2j + uL1j uL2j

j=1 w2j 1, 2, . . . , m) and S ∗ is obtained by using one of Equa-

+uU1j uU2j + v L1j v L2j + v U1j v U2j tions (15), (16), (23) and (24) to rank the alternatives.

= ⎛ L 2 U 2

⎞.

n (t1j ) + (t1j ) + (uL1j )2 Thus, the greater the value of the weighted gener-

⎜λ w2j + ⎟ alized Dice measure between Si (i = 1, 2, . . . , m)

⎜ j=1

+(uU1j )2 + (v L1j )2 + (v U1j )2 ⎟

⎜ ⎟ and S ∗ is, the better the alternative Si is.

⎜ ⎟

⎜ n L 2

(t2j U 2

) + (t2j ) + (uL2j )2 ⎟

⎝ ⎠

(1 − λ) w2j

j=1

+(uU2j )2 + (v L2j )2 + (v U2j )2 6. Decision-making example

(24) of manufacturing schemes

5. Decision making-methods based lem of manufacturing schemes with simplified

on the generalized Dice measures neutrosophic information is given to demonstrate

the applications and effectiveness of the proposed

In this section, we propose multiple attribute decision-making methods in realistic scenarios.

decision-making methods by using the generalized To select the best manufacturing scheme (alter-

Dice measures of SNSs under simplified neutro- native) for the flexible manufacturing system in a

sophic environment. manufacturing company, the technique department of

For multiple attribute decision-making problems, the company provides four manufacturing schemes

let S = {S1 , S2 , . . . , Sm } be a set of alternatives and (alternatives) with respect to some product as a set

R = {R1 , R2 , . . . , Rn } be a set of attributes. Then, of the alternatives S = {S1 , S2 , S3 , S4 } for the flex-

the weight of the attribute

Rj (j = 1, 2, . . . , n) is ible manufacturing system. A decision must be made

wj , wj ∈ [0, 1] and nj=1 wj = 1. Thus, the fit judg- according to the four attributes: (1) R1 is the improve-

ment (satisfaction evaluation) of an attribute Rj (j = ment of quality; (2) R2 is the market response; (3)

1, 2, . . . , n) for an alternative Si (i = 1, 2, . . . , m) R3 is the manufacturing cost; (4) R4 is the manu-

is represented by a SNS Si = {si1 , si2 , . . . , sin }, facturing complexity. The weight vector of the four

where sij =< tij , uij , v ij > is a SVNN for 0 ≤ tij + attributes W = (0.3, 0.25, 0.25, 0.2)T is given by

uij + v ij ≤ 3 or an INN for 0 ≤ tijU + uU U

ij + v ij ≤ decision makers.

3 (j = 1, 2, . . . , n and i = 1, 2, . . . , m). Therefore, In the decision-making problem, the decision mak-

we can establish a simplified neutrosophic decision ers are required to make the fit judgment (satisfaction

evaluation) of an attribute Rj (j = 1, 2, 3, 4) for an

matrix D = (sij )m×n .

alternative Si (i = 1, 2, 3, 4) and to give simplified

In the multiple attribute decision-making problem, neutrosophic evaluation information, which is shown

the concept of an ideal solution (ideal alternative) can in the following decision matrix with SVNNs:

J. Ye / The generalized Dice measures for multiple attribute decision making 669

D= Table 1

⎡ ⎤ The measure values of Equation (15) and ranking orders

(0.75, 0.2, 0.3) (0.7, 0.2, 0.3) (0.65, 0.2, 0.25) (0.75, 0.2, 0.1)

λ GWSVNN1 GWSVNN1 GWSVNN1 GWSVNN1 Ranking order

⎢ (0.8, 0.1, 0.2) (0.75, 0.2, 0.1) (0.75, 0.2, 0.1) (0.85, 0.1, 0.2) ⎥

(S1 , S ∗ ) (S2 , S ∗ ) (S3 , S ∗ ) (S4 , S ∗ )

⎢ ⎥.

⎣ (0.7, 0.2, 0.2) (0.78, 0.2, 0.1) (0.85, 0.15, 0.1) (0.76, 0.2, 0.2) ⎦ 0 0.8895 0.9517 0.9361 0.9287 S2

S3

S4

S1

(0.8, 0.2, 0.1) (0.85, 0.2, 0.2) (0.7, 0.2, 0.2) (0.86, 0.1, 0.2) 0.2 0.9157 0.9667 0.9558 0.9472 S2

S3

S4

S1

0.5 0.9612 0.9924 0.9876 0.9816 S2

S3

S4

S1

Then, the developed decision-making methods can 0.7 0.9966 1.0119 1.0104 1.0100 S2

S3

S4

S1

be used for the decision making problem. 1 1.0594 1.0455 1.0475 1.0641 S4

S1

S3

S2

According to

sj∗ =< tj∗ , u∗j , v ∗j >= < max(tij ), min(uij ), min(v ij ) >

Table 2

i i i The measure values of Equation (16) and ranking orders

for j = 1, 2, 3, 4 and i = 1, 2, 3, 4, we can obtain λ GWSVNN2 GWSVNN2 GWSVNN2 GWSVNN2 Ranking order

(S1 , S ∗ ) (S2 , S ∗ ) (S3 , S ∗ ) (S4 , S ∗ )

an ideal solution (ideal alternative) as follows:

0 0.8908 0.9503 0.9387 0.9372 S2

S3

S4

S1

S ∗ = {s1∗ , s2∗ , s3∗ , s4∗ } 0.2 0.9175 0.9667 0.9587 0.9557 S2

S3

S4

S1

0.5 0.9605 0.9924 0.9903 0.9849 S2

S3

S4

S1

< 0.8, 0.1, 0.1 >, < 0.85, 0.2, 0.1 >, 0.7 0.9915 1.0102 1.0126 1.0054 S3

S2

S4

S1

= . 1 1.0419 1.0383 1.0479 1.0378 S3

S1

S2

S4

< 0.85, 0.15, 0.1 >, < 0.86, 0.1, 0.1 >

By using Equation (15) or (16) and different values the alternative S2 is the best choice among all

of the parameter λ, the weighted generalized Dice the alternatives.

measure values between Si (i = 1, 2, 3, 4) and S ∗ (3) When λ = 1, the two weighted generalized

can be obtained, which are shown in Tables 1 and 2 Dice measures are reduced to the weighted

respectively. projection measures of S ∗ on Si . Thus, the

From Tables 1 and 2, we can see that different rank- alternative S3 or S4 is the best choice among

ing orders are indicated by taking different values of all the alternatives.

the parameter λ and different generalized Dice mea- Obviously, according to different values of the

sures. Then we can obtain that the best alternative is parameter λ and different measures, ranking orders

S2 or S3 or S4 . may be different. Thus the proposed decision-making

Furthermore, for the special cases of the two gener- methods can be assigned some value of λ and some

alized Dice measures we obtain the following results: measure to satisfy the decision makers’ preference

(1) When λ = 0, the two weighted generalized and/or real requirements.

Dice measures are reduced to the weighted If the fit judgment (satisfaction evaluation) of an

projection measures of Si on S ∗ . Thus, the attribute Rj (j = 1, 2, 3, 4) for an alternative Si

alternative S2 is the best choice among all the (i = 1, 2, 3, 4) is given in the decision making

alternatives. problem by the following decision matrix with INNs:

D=

⎡

< [0.7, 0.8], [0.1, 0.2], [0.2, 0.3] > < [0.7, 0.8], [0.1, 0.2], [0.2, 0.3] >

⎢

⎢ < [0.7, 0.9], [0.1, 0.2], [0.2, 0.3] > < [0.7, 0.8], [0.1, 0.3], [0.1, 0.2] >

⎢

⎢ < [0.7, 0.8], [0.1, 0.3], [0.2, 0.3] > < [0.8, 0.9], [0.1, 0.2], [0.1, 0.2] >

⎣

< [0.8, 0.9], [0.2, 0.3], [0.1, 0.2] > < [0.8, 0.9], [0.2, 0.3], [0.2, 0.3] >

⎤

< [0.6, 0.7], [0.1, 0.2], [0.2, 0.4] > < [0.7, 0.8], [0.1, 0.2], [0.1, 0.2] >

⎥

< [0.7, 0.8], [0.2, 0.3], [0.1, 0.2] > < [0.8, 0.9], [0.0, 0.1], [0.2, 0.3] > ⎥

⎥.

< [0.8, 0.9], [0.1, 0.2], [0.1, 0.1] > < [0.7, 0.8], [0.1, 0.2], [0.1, 0.2] > ⎥

⎦

< [0.7, 0.8], [0.1, 0.2], [0.1, 0.3] > < [0.7, 0.9], [0.0, 0.1], [0.1, 0.3] >

Dice measures are reduced to the weighted

Dice similarity measures of Si and S ∗ . Thus,

670 J. Ye / The generalized Dice measures for multiple attribute decision making

The measure values of Equation (23) and ranking orders

sj∗ =< tj∗ , u∗j , v ∗j >=< [max(tijL ), max(tijU )], λ GWINN3 GWINN3 GWINN3 GWINN3 Ranking order

i i

(S1 , S ∗ ) (S2 , S ∗ ) (S3 , S ∗ ) (S4 , S ∗ )

[min(uL U L U

ij ), min(uij )], [min(v ij ), min(v ij )] > 0 0.9085 0.9770 0.9861 1.0159 S4

S3

S2

S1

i i i i 0.2 0.9325 0.9819 0.9860 1.0015 S4

S3

S2

S1

0.5 0.9737 0.9903 0.9901 0.9863 S2

S3

S4

S1

for j = 1, 2, 3, 4 and j = 1, 2, 3, 4, we can

0.7 1.0053 0.9966 0.9955 0.9797 S1

S2

S3

S4

obtain the ideal solution (ideal alternative) as follows: 1 1.0601 1.0072 1.0075 0.9746 S1

S3

S2

S4

⎧ ⎫

⎪

⎪ < [0.8, 0.9], [0.1, 0.2], [0.1, 0.2] >,⎪

⎪

Table 4

⎪

⎪ ⎪

⎪ The measure values of Equation (24) and ranking orders

⎨ < [0.7, 0.8], [0.1, 0.2], [0.1, 0.2] >,⎪

⎪ ⎬ λ GWINN4 GWINN4 GWINN4 GWINN4 Ranking order

= . (S1 , S ∗ ) (S2 , S ∗ ) (S3 , S ∗ ) (S4 , S ∗ )

⎪

⎪ < [0.8, 0.9], [0.1, 0.2], [0.1, 0.1] >,⎪

⎪

⎪

⎪ ⎪

⎪ S4

S3

S2

S1

⎪

⎩ ⎪

⎭

0 0.9035 0.9726 0.9790 1.0109

< [0.8, 0.9], [0.0, 0.1], [0.1, 0.2] > 0.2 0.9306 0.9795 0.9833 1.0011 S4

S3

S2

S1

0.5 0.9743 0.9901 0.9900 0.9867 S2

S3

S4

S1

According to Equations (23) or (24) and different 0.7 1.0059 0.9972 0.9945 0.9774 S1

S2

S3

S4

1 1.0572 1.0082 1.0013 0.9637 S1

S2

S3

S4

values of the parameter λ, the weighted generalized

Dice measure values between Si (i = 1, 2, 3, 4) and

S ∗ can be obtained, which are shown in Tables 3 and methods based on generalized Dice measures.

4 respectively. Therefore, in the decision-making process, the

From Tables 3 and 4, different ranking orders are decision-making methods developed in this paper

shown by taking different values of λ and different are more general and more flexible than existing

measures. Then we can obtain that the best alternative decision-making methods under simplified neutro-

is S1 or S2 or S4 . sophic environment.

Furthermore, for the special cases of the two gener-

alized Dice measures we obtain the following results:

7. Conclusion

(1) When λ = 0, the two weighted generalized

Dice measures are reduced to the weighted

This paper proposed another form of the Dice

projection measures of Si on S ∗ . Thus, the

measures between SNSs and the generalized Dice

alternative S4 is the best choice among all the

measures of SNSs and indicated the Dice measures

alternatives.

of SNSs and the projection measures (asymmetric

(2) When λ = 0.5, the two weighted generalized

measures) of SNSs are the special cases of the gen-

Dice measures are reduced to the weighted

eralized Dice measures of SNSs corresponding to

Dice similarity measures of Si and S ∗ . Thus,

some parameter values. Then, we developed multi-

the alternative S2 is the best choice among all

ple attribute decision-making methods based on the

the alternatives.

generalized Dice measures of SNSs under simplified

(3) When λ = 1, the two weighted generalized

neutrosophic environment. According to different

Dice measures are reduced to the weighted

parameter values and some measure preferred by

projection measures of S ∗ on Si . Thus, the

decision makers, by the weighted generalized Dice

alternative S1 is the best choice among all the

measure between each alternative and the ideal

alternatives.

solution (ideal alternative), all alternatives can be

Therefore, according to different values of the ranked and the best alternative can be selected as

parameter λ and different measures, ranking orders well. Finally, a real example about the selection

may be also different. Thus the proposed decision- of manufacturing schemes (alternatives) demon-

making methods can be assigned some value of λ strated the applications of the developed methods

and some measure to satisfy the decision makers’ under simplified neutrosophic environment, and then

preference and/or real requirements. the effectiveness and flexibility of the developed

Obviously, the decision-making methods based decision-making methods were shown corresponding

the Dice measures and the projection measures are to different parameter values. In the decision-making

the special cases of the proposed decision-making process under simplified neutrosophic environment,

J. Ye / The generalized Dice measures for multiple attribute decision making 671

the main advantage is more general and more flexible [11] J. Ye, Similarity measures between interval neutrosophic

than existing decision-making methods to satisfy the sets and their applications in multicriteria decision-making,

Journal of Intelligent and Fuzzy Systems 26 (2014),

decision makers’ preference and/or practical require- 165–172.

ments. [12] J. Ye, Multiple attribute group decision-making method with

In the future work, we shall extend the generalized completely unknown weights based on similarity measures

Dice measures of SNSs to other areas such as pattern under single valued neutrosophic environment, Journal of

Intelligent and Fuzzy Systems 27(12) (2014), 2927–2935.

recognition, fault diagnosis, and image processing. [13] J. Ye, Vector similarity measures of simplified neutrosophic

sets and their application in multicriteria decision mak-

ing, International Journal of Fuzzy Systems 16(2) (2014),

References 204–211.

[14] J. Ye, Single valued neutrosophic cross-entropy for multi-

criteria decision making problems, Applied Mathematical

[1] F. Smarandache, Neutrosophy. Neutrosophic Probability, Modelling 38(3) (2014), 1170–1175.

Set, and Logic, American Research Press, Rehoboth, USA [15] J. Ye, Multiple attribute decision-making method based on

1998. the possibility degree ranking method and ordered weighted

[2] G.L. Xu and F. Liu, An approach to group decision making aggregation operators of interval neutrosophic numbers,

based on interval multiplicative and fuzzy preference rela- Journal of Intelligent and Fuzzy Systems 28(3) (2015),

tions by using projection, Applied Mathematical Modelling 1307–1317.

37 (2013), 3929–3943. [16] K. Atanassov, Intuitionistic fuzzy sets, Fuzzy Sets and

[3] H. Wang, F. Smarandache, Y.Q. Zhang and R. Sunderraman, Systems 20 (1986), 87–96.

Interval neutrosophic sets and logic: Theory and applica- [17] K. Atanassov and G. Gargov, Interval valued intuitionistic

tions in computing, Hexis, Phoenix, A.Z. 2005. fuzzy sets, Fuzzy Sets and Systems 31 (1989), 343–349.

[4] H. Wang, F. Smarandache, Y.Q. Zhang and R. Sunderraman, [18] P.P. Chi and P.D. Liu, An Extended TOPSIS Method for

Single valued neutrosophic sets, Multispace and Multistruc- multiple attribute decision making problems based on inter-

ture 4 (2010), 410–413. val neutrosophic set, Neutrosophic Sets and Systems 1

[5] H.Y. Zhang, J.Q. Wang and X.H. Chen, Interval neutro- (2013), 63–70.

sophic sets and their application in multicriteria decision [19] P.D. Liu, Y.C. Chu, Y.W. Li and Y.B. Chen, Some general-

making problems, Science World Journal 2014 (2014), 15. ized neutrosophic number Hamacher aggregation operators

Article ID 645953. and their application to group decision making, Interna-

[6] H.Y. Zhang, J.Q. Wang and X.H. Chen, An outranking tional Journal of Fuzzy Systems 16(2) (2014), 242–255.

approach for multi-criteria decision-making problems with [20] P.D. Liu and Y.M. Wang, Multiple attribute decision mak-

interval-valued neutrosophic sets, Neural Computing and ing method based on single valued neutrosophic normalized

Applications (2015). doi: 10.1007/s00521-015-1882-3 weighted Bonferroni mean, Neural Computing and Appli-

[7] J.J. Peng, J.Q. Wang, H.Y. Zhang and X.H. Chen, An cations 25(7-8) (2014), 2001–2010.

outranking approach for multi-criteria decision-making [21] P. Biswas, S. Pramanik and B.C. Giri, TOPSIS method for

problems with simplified neutrosophic sets, Applied Soft multi-attribute group decision-making under single-valued

Computing 25 (2014), 336–346. neutrosophic environment, Neural Computing and Applica-

[8] J.J. Peng, J.Q. Wang, J. Wang, H.Y. Zhang and X.H. tions (2015). doi: 10.1007/s00521-015-1891-2

Chen, Simplified neutrosophic sets and their applica- [22] R. Sahin and A. Kucuk, Subsethood measure for single

tions in multi-criteria group decision-making problems, valued neutrosophic sets, Journal of Intelligent and Fuzzy

International Journal of Systems Science (2015). doi: Systems 29(2) (2015), 525–530.

10.1080/00207721.2014.994050 [23] R. Şahin and P.D. Liu, Maximizing deviation method

[9] J. Ye, Multicriteria decision-making method using the for neutrosophic multiple attribute decision making with

correlation coefficient under single-valued neutrosophic incomplete weight information, Neural Computing and

environment, International Journal of General Systems Applications (2015). doi: 10.1007/s00521-015-1995-8

42(4) (2013), 386–394.

[10] J. Ye, A multicriteria decision-making method using aggre-

gation operators for simplified neutrosophic sets, Journal of

Intelligent and Fuzzy Systems 26 (2014), 2459–2466.

Copyright of Journal of Intelligent & Fuzzy Systems is the property of IOS Press and its

content may not be copied or emailed to multiple sites or posted to a listserv without the

copyright holder's express written permission. However, users may print, download, or email

articles for individual use.

## Molto più che documenti.

Scopri tutto ciò che Scribd ha da offrire, inclusi libri e audiolibri dei maggiori editori.

Annulla in qualsiasi momento.