Distance and Similarity Measures For Bipolar Fuzzy Soft Sets With Application To Pharmaceutical Logistics and Supply Chain Management

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/358554763

Distance and similarity measures for bipolar fuzzy soft sets with application
to pharmaceutical logistics and supply chain management

Article in Journal of Intelligent and Fuzzy Systems · February 2022


DOI: 10.3233/JIFS-210873

CITATIONS READS

7 289

4 authors:

Muhammad Riaz Mishal Riaz


University of the Punjab University of the Punjab
156 PUBLICATIONS 3,126 CITATIONS 1 PUBLICATION 7 CITATIONS

SEE PROFILE SEE PROFILE

Nimra Jamil Zarife Zararsız


University of the Punjab Nevşehir Hacı Bektaş Veli University
7 PUBLICATIONS 26 CITATIONS 22 PUBLICATIONS 113 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Pythagorean m-Polar Fuzzy Sets View project

Linear Diophantine Fuzzy Sets View project

All content following this page was uploaded by Muhammad Riaz on 05 July 2022.

The user has requested enhancement of the downloaded file.


Journal of Intelligent & Fuzzy Systems 42 (2022) 3169–3188 3169
DOI:10.3233/JIFS-210873
IOS Press

Distance and similarity measures for bipolar


fuzzy soft sets with application to
pharmaceutical logistics and supply chain
management

PY
Muhammad Riaza,∗ , Mishal Riaza , Nimra Jamila and Zarife Zararsizb
a Department of Mathematics, University of the Punjab, Lahore, Pakistan

CO
b Department of Mathematics, Nevsehir Hacı Bektas¸ Veli University, Nevsehir, Turkey

Abstract. Pharmaceutical logistics are primarily concerned with handling transportation and supply chain management of
OR
numerous complex goods most of which need particular requirements for their logistical care. To find the high level of
specialization, suppliers of pharmaceutical logistics must be selected under a mathematical model that can treat vague and
uncertain real-life circumstances. The notion of bipolarity is a key factor to address such uncertainties. A bipolar fuzzy soft set
(BFSS) is a strong mathematical tool to cope with uncertainty and unreliability in various real-life problems including logistics
and supply chain management. In this paper, we introduce new similarity measures (SMs) based on certain properties of
bipolar fuzzy soft sets (BFSSs). The proposed SMs are the extensions of Frobenius inner product, cosine similarity measure,
TH

and weighted similarity measure for BFSSs. The proposed SMs are also illustrated with respective numerical examples. An
innovative multi-attribute decision-making algorithm (MADM) and its flow chart are being developed for pharmaceutical
logistics and supply chain management in COVID-19. Furthermore, the application of the suggested MADM method is
presented for the selection of the best pharmaceutical logistic company and a comparative analysis of the suggested SMs
with some of the existing SMs is also demonstrated.
AU

Keywords: Bipolar fuzzy soft sets, similarity measures, pharmaceutical logistics, multi-attribute decision-making

1. Introduction between the two entities is proportionate to the value


of the SM.
The principle of similarity measurement (SM) For instance, SM may be used in a business case
is more or less essential in any field of creativity, firm to match resumes with the job description, saving
science, and technology. It is usually carved to deter- the recruiter a considerable amount of time. Similarly,
mine the trustworthiness of a particular instance, by finding out SM between the symptoms present in a
person, product, object, or system. SM is an important patient in medical encounter and the symptoms of that
method of assessing the degree of similarity between particular disease suspected for the patient, we may
two or more information systems. The similarity diagnose the patient with a better degree of treatment.
In particular cases, the selection method for a specific
∗ Corresponding author. Muhammad Riaz, Department of logistics supplier can reduce the cost and time factor
Mathematics, University of the Punjab, Lahore, Pakistan. E-mail: and increase the safety and satisfaction of logistic
mriaz.math@pu.edu.pk. care.

ISSN 1064-1246/$35.00 © 2022 – IOS Press. All rights reserved.


3170 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

Zadeh [1] is the foremost researcher who created fuzzy soft mappings with decision-making in medical
the concept of the fuzzy set (FS) to address vagueness diagnosis and bipolar disorder.
and uncertainty. A number of theories and mod- Alcantud et al. [31] proposed the notion of dual
els have been developed for modeling uncertainties extended hesitant fuzzy sets and their related results.
including, intuitionistic fuzzy set (IFS) [2, 3], soft set Zhang and Xu [32] proposed the idea of Pythagorean
(SS) [4], bipolar fuzzy set (BFS) [5, 6], Pythagorean fuzzy numbers (PyFNs) and their fundamental oper-
fuzzy set (PyFS) [7–9], q-rung orthopair fuzzy set (q- ations. They proposed MCDM approach with a new
ROFS) [10], neutrosophic set (NS) [11], hypersoft set extension of TOPSIS based on PyFSs. They devel-
and plithogenic hypersoft set [12], spherical fuzzy set oped an application of proposed MCDM approach to
(SFS) [13–16], etc. examine the service quality among domestic airlines.
Molodtsov [4] introduced an abstraction of crisp Decision-making with similarity measures and
set, namely soft set as a robust model that treats ambi- distance measures have been studied by many

PY
guities and uncertainties in the parametric ways. The researchers; for soft sets (SSs) (see [33–37]), for
parameters are evaluated in terms of crisp subsets or various fuzzy sets and vague sets (see [38–41]), for
classes by a set-valued mapping which is an essential IFSs (see [42–48]), for PyFSs (see [49–54]), for q-
component in evaluating the objects in the decision rung orthopair fuzzy sets (qROFS) (see [55–58]), for
analysis. In short, the soft set theory has fascinatingly bipolar fuzzy set and bipolar neutrosophic set (see

CO
solved many real-world problems in the implementa- [59–61]), and soft sets and their algebraic structures
tion of current hypotheses. In theoretical and science (see [62, 66]. Zararsiz [67, 68] proposed similarity
emphasis, soft set theory has gained great popularity. measures of sequence of fuzzy numbers, algebraic
Bipolarity of information is a critical factor in structure of fuzzy numbers, and fuzzy risk analysis.
the analysis of data in a number of forms, part of Zararsiz and sengonul [69] studied the gravity of cen-
which must be taken into account when constructing ter of sequence of fuzzy numbers. However, distance
a mathematical structure for many of the scenarios. and similarity measures for bipolar fuzzy soft sets
OR
Bipolarity reflects the positive and negative aspects have not been studied by the researchers. To fill this
of a particular topic. For instance, two distinct aspects research gap, we introduce some new distance and
of decision analysis are happiness and sadness, sweet similarity measures for bipolar fuzzy soft sets, and
and sour, effects and side effects. Zhang [5, 6] intro- then develop an algorithm for modeling uncertainties
duced the notion of bipolarity in both crisp and fuzzy in decision-making.
TH

sets, namely bipolar sets and bipolar fuzzy set (BFS), The remaining paper is arranged as follows: In Sec-
respectively. Wei et al. [17] proposed the idea of tion 2, some basics concepts of BFSs and BFSSs
interval-valued bipolar fuzzy set (IVBFS) and dis- are recalled. Distance and similarity measures of
cussed the multi-criteria decision-making (MCDM) BFSSs are defined in Section 3. Weighted similar-
for emerging technology commercialization evalua- ity measures for BFSSs are proposed in Section 4.
AU

tion. The abstractions of bipolar fuzzy set and bipolar A robust application to pharmaceutical logistics and
fuzzy soft set have been studied in the last decades supply chain management is presented in Section 5.
by many researchers (see [18–22]). Lastly, the summary of research work is concluded in
Lee [23, 24] studied bipolar-valued fuzzy set Section 6.
(BVFS), their basic operations, and described the
comparison of interval-valued fuzzy sets (IVFSs),
IFSs, and BVFSs. Batyrshin [25, 26] proposed a gen- 2. Preliminaries
eral theory of similarity and association measures,
dissimilarity, resemblance functions, and correla- In this section, we discuss basic concepts of soft
tion functions. These concepts play an important sets and bipolar fuzzy sets. A detailed study of bipolar
role in information retrieval and information mea- fuzzy sets and bipolar fuzzy soft sets can be seen in
sures, statistics, data science, recommender systems, [5, 6, 19, 20, 21, 23, 24, 27].
machine learning, and decision-making. Akram [27]
proposed the novel concept of bipolar fuzzy graphs Definition 2.1. [4] Let M be the universe, P be the
and their application. Alghamdi et al. [28] proposed set of parameters, and A ⊆ P. Then, a soft set ϒA
new MCDM approach for bipolar fuzzy environment. over M is defined by
Riaz and Tehrim [29, 30] proposed the notion of bipo-
lar fuzzy soft topology (BFS-topology) and bipolar ϒA = {(p, ϒ(p)) : p ∈ A, ϒ(p) ∈ (M)}
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3171

where (M) is the power set of M, and ϒ maps A to Table 1


Bipolar fuzzy soft set A
(M).
A p1 p2 ··· pn
Definition 2.2. [5] Let M be the universe. A bipolar 1 (μ+ −
11 , μ11 ) (μ+ −
12 , μ12 ) ··· (μ+ −
1n , μ1n )
fuzzy set (BFS) on M is defined by 2 (μ+
21 , μ −
21 ) (μ+
22 , μ −
22 ) ··· (μ+
2n , μ −
2n )
.. .. .. .. ..
  . . . . .
B = (, μ+ −
B (), μB ()): for all  ∈ M m (μ+ −
m1 , μm1 ) (μ+ −
m2 , μm2 ) ··· (μ+ −
mn , μmn )

where μ+ −
B () ∈ [0, 1] and μB () ∈ [−1, 0] represent
the positive membership and negative membership, Definition 2.4. [19] Let M be the universe of dis-
respectively. course, P the set of parameters, and A ⊆ P. Let
The collection of all BFSs on M can be expressed as  : A → BF M be a mapping, then a bipolar fuzzy
BF (M) or BF M . soft set (BFSS) (, A) or A is defined by

PY
  
Definition 2.3. [23] Let B, B1 and B2 ∈ BF (M) be (, A) = p, (p) : p ∈ A, (p) ∈ BF M

BFSs. Then,
  
1. Null BFS: A BFS B ∈ BF (M) is said to be null + −
= p, {, μB (), μB ()} : p ∈ A,  ∈ M

CO
BFS, if μ+ −
B (i ) = 0 and μB (i ) = 0,
 It
for each i ∈ M and we denote it as φB or φ. If M = {1 , · · · , m }, A = {p1 , · · · , pn }, then BFSS
can be expressed as follows: A in the tabular form is expressed in Table 1 . and
  corresponding bipolar fuzzy soft matrix (shortly BFS
φB = (, 0, 0) :  ∈ M matrix) is given by
2. Absolute BFS: A BFS B ∈ BF (M) is called an (, A) = [(μ+ −
OR
ij , μij )]m×n
absolute BFS, if μ+ − ⎛ ⎞
B (i ) = 1 and μB (i ) = −1, (μ+ −
(μ+ −
··· (μ+ −
11 , μ11 ) 12 , μ12 ) 1n , μ1n )

for each i ∈ M and we denote it as MB or M.It ⎜ (μ+21 , μ−21 ) − ⎟
⎜ (μ+ −
22 , μ22 ) ··· (μ+
2n , μ2n ) ⎟
can be expressed as follows: =⎜
⎜ .. .. .. .. ⎟.

  ⎝ . . . . ⎠
MB = (, 1, −1) :  ∈ M (μ+ −
(μ+ −
··· (μ+ −
m1 , μm1 ) m2 , μm2 ) mn , μmn )
TH

3. Complement: The compliment of BFS, B ∈


BF (M) is defined by
Definition 2.5. [19] A BFSS A ∈ BF M is called a
B = Bc = {(i , 1 − μ+ − null BFSS, if μ+ −
A () = 0 and μA () = 0, for each
B (i ), −1 − μB (i )) :
 ∈ M. It can be expressed as follows:
AU

i ∈ M}  

φA = p, {, 0, 0} : p ∈ P,  ∈ M
4. Inclusion: Consider two BFS B1 and B2 ∈
BF (M), then B1 ⊆ B2 , if and only if
μ+ + − −
B1 (i ) ≤ μB2 (i ) and μB1 (i ) ≥ μB2 (i ). Definition 2.6. [19] A BFSS A ∈ BF M is called
5. Intersection: Let B1 and B2 ∈ BF (M), then the
an absolute BFSS, if μ+ −
C () = 1 and μC = −1, for
intersection of B1 and B2 is a BFS given by
each  ∈ M. It can be expressed as follows:

B1 ∩ B2 = i , min(μ+ +   
B1 (i ), μB2 (i )), 
M A = p, {, 1, −1} : p ∈ P,  ∈ M

max(μ− B1 ( i ), μ−
B2 ( i )) .
(1) (2)
6. Union: Let B1 and B2 ∈ BF (M), then the union Definition 2.7. [19] Let A1 = (1 , A1 ) and A2 =
of B1 and B2 is a BFS given by (2 , A2 ) be BFSSs over M, where A1 , A2 ⊆ P.
(1) (2) (1)  (2)
 Then, A1 is BFS-subset of A2 i.e. A1 ⊆ A2 , if
B1 ∪ B2 = i , max(μ+ +
B1 (i ), μB2 (i )),
the following situations hold
 (i) A1 ⊆ A2
min(μ− −
B1 (i ), μB2 (i )) . (ii) 1 (p) is BF-subset of 2 (p) for all p ∈ A1 .
3172 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

(1) (2) In short, it can be written as follows:


Two BFSSs A1 and A2 over M are said
(1) (2) 
to be equal, written A1 = A2 , if and only if A = p2 , {(2 , 0.42, −0.79), (3 , 0.72, −0.21),
(1)  (2)  (1)
A ⊆ A ⊆A hold. 
1 2 1
(4 , 0.21, −0.59)} ,


Definition 2.8. [19] Let (1 , A1 ) and (2 , p4 , {(1 , 0.59, −0.41), (3 , 0.92, −0.11),
A2 ) be two BFSSs over M. Then, (, A) = 
 (2 , A2 ) , where A = A1 ∪ A2 and for all
(1 , A1 ) ∪ (4 , 0.69, −0.23)} .
p ∈ A,
⎧ Then, corresponding BFS matrix is obtained as fol-

⎨ 1 (p) , if p ∈ A1 \A2 lows:
 (p) = 2 (p) , if p ∈ A2 \A1

PY

⎩ (, A)
1 (p) ∪ 2 (p) , if p ∈ A1 ∩ A2 ⎛ ⎞
(0, 0) (0, 0) (0, 0) (0.59, −0.41)
where 1 (p) ∪ 2 (p) is the union of two BFSs. ⎜(0, 0) (0.42, −0.79) (0, 0) ⎟
(0, 0)
⎜ ⎟
=⎜ ⎟.
⎝(0, 0) (0.72, −0.21) (0, 0) (0.92, −0.11)⎠
Definition 2.9. [19] The intersection of two BFSSs

CO
(0, 0) (0.21, −0.59) (0, 0) (0.69, −0.23)
(1 , A1 ) and (2 , A2 ) is a BFS-set (, A) =
 (2 , A2 ), where A = A1 ∩ A2 =
(1 , A1 ) ∩ / φ and
 (p) = 1 (p) ∩ 2 (p) for all p ∈ A.
Definition 2.12. Let B1 , B2 be two BFSs. A measure
Definition 2.10. [19] The complement of the BFSS Sim(B1 , B2 ) is termed as a similarity measure (SM)
    if it fulfill the following requirements:
A = p, , μ+ −
: p ∈ A,  ∈ M
OR
A (), μA ()
1. 0 ≤ Sim(B1 , B2 ) ≤ 1.
2. Sim(B1 , B2 ) = 1 ⇔ B1 = B2 .
is 3. Sim(B1 , B2 ) = Sim(B2 , B1 ).
 
A = cA = p, , 1 − μ+
A (), −1 Definition 2.13. [19] Let (ϒ1 , P1 ) and (ϒ2 , P2 ) be
TH

  two soft sets. If P1 and P2 are coincident (same set


−μ−
A () : p ∈ A,  ∈ M . of parameters), then the similarity between (ϒ1 , P1 )
and (ϒ2 , P2 ) is expressed as
i F1 (pi ).F2 (pi )
Example 2.11. Let M = {1 , 2 , 3 , 4 } be the S(ϒ1 , ϒ2 ) = .
AU

i F1 (pi )2 ∨ F2 (pi )2
set of four juicer machines under consideration
and P = {p1 = High Speed, p2 = Efficient, p3 = We compute S(ϒ1 , ϒ2 ) utilizing above formula if P1
Modern Technology, p4 = Warranty} such that and P2 are overlapping and different, by consider-
A = {p2 , p3 } ⊆ P. Then, the BFSS over M cab be ing first definition F1 (p) = ∅ for p ∈ P2 \ (P1 ∩ P2 )
written as follows: and F2 (p) = ∅ for p ∈ P1 \ (P1 ∩ P2 ), where ∅ is the
  empty set.
A = p1 , {(1 , 0, 0), (2 , 0, 0), (3 , 0, 0), (4 , 0, 0)} ,
Definition 2.14. [35] Let M = {1 , · · · , n } be the

p2 , {(1 , 0, 0), (2 , 0.42, −0.79), underlying set and wi ∈ [0, 1] (not all zero) be the
 weight of i . The weighted similarity of two soft sets
(3 , 0.72, −0.21), (4 , 0.21, −0.59)} , (ϒ1 , P) and (ϒ2 , P) defined over M is
  1 i wi (ϒ1 (pi ).ϒ2 (pi ))
p3 , {(1 , 0, 0), (2 , 0, 0), (3 , 0, 0), (4 , 0, 0)} , W(ϒ1 , ϒ2 ) = × .
i w i i ϒ1 (pi )2 ∨ ϒ2 (pi )2

p4 , {(1 , 0.59, −0.41), (2 , 0, 0),
 Definition 2.15. [36] Let M = {i : i = 1, · · · , m}
(3 , 0.92, −0.11), (4 , 0.69, −0.23)} . be a crisp set with collection of attributes P =
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3173

{pi : i = 1, 2, · · · , n}. If P1 , P2 ⊆ P, then the SM between them is given by


between the intuitionistic fuzzy soft sets (IFSSs) i G1 (pi ).G2 (pi )
(N, P1 ) and (K, P2 ) is given by S(G1 , G2 ) =
i max[G1 (pi )2 , G2 (pi )2 ]
S(N, K)
Note that G1 (p) = 0 for p ∈ P2 \ (P1 ∩ P2 ) and

→ −
→+ −
→− −
→− G2 (p ) = 0 for p ∈ P1 \ (P1 ∩ P2 ).
i [ N + μ (pi ) K μ (pi ) + N μ (pi ) K μ (pi )]
= −
→ −
→+ −
→− −
→− .
μ (pi ) ∨ K μ (pi ) } + { N μ (pi ) ∨ K μ (pi ) }]
i [{ N + 2 2 2 2
Definition 3.3. [35] Let M = {1 , 2 , 3 , · · · } be a
universe, P = {p1 , p2 , p3 , · · · } be the set of parame-
ters, A1 , A2 ⊆ P and ϒA1 , ϒA2 be two soft sets on M
with their approximate function fA1 and kA2 , respec-
3. Distance and similarity measures of BFSSs tively. Then, the distance of ϒA1 and ϒA2 are defined

PY
as,
In this section, we present distance and similarity
1. Hamming distance:
measures for BFSSs. For distance measures between
bipolar fuzzy sets and soft sets the readers can see d 1 (ϒA1 , ϒA2 )
[18, 28, 35]. Throughout this manuscript, M is the  m n 
1 

CO
universe of discourse and P is the set of parameters.
= |fA1 (pi )(j ) − kA2 (pi )(j )|
m
i=1 j=1
Definition 3.1. [18, 28]
Let M = {1 , 2 , · · · , n }, B1 and B2 be two BFSs 2. Normalized Hamming distance:
over M with their positive membership function μ+ B1 d 2 (ϒA1 , ϒA2 )
and μ+ and negative membership function μ −
and
B2 B1  m n 
μ−B2 respectively. Then, Han et al. [18] defined the
1 
OR
= |fA1 (pi )(j ) − kA2 (pi )(j )|
distances between B1 and B2 as follows: mn
i=1 j=1
1. Hamming Distance: 3. Euclidean distance:
1  +
n
d 3 (ϒA1 , ϒA2 )
dh (B1 , B2 ) = (|μB1 (i ) − μ+
B2 (i )| 
TH

2n  
i=1 1 m  n
= (fA1 (pi )(j ) − kA2 (pi )(j ))2
+ |μ− −
B1 (i ) − μB2 (i )|) m
i=1 j=1
2. Poles-weighted Hamming distance: 4. Normalized Euclidean distance:
AU

1 p +
n
d 4 (ϒA1 , ϒA2 )
dph (B1 , B2 ) = (κ |μB1 (i ) − μ+
B2 (i )|
n 

i=1
 1  m  n

+κ |μ− − = (fA1 (pi )(j ) − kA2 (pi )(j ))2


B1 (i ) − μB2 (i )|)
N
mn
i=1 j=1
where κp + κN
= 1.
3. Alghamdi et al. [28] defined Euclidean distance Definition 3.4. [35] Let ϒA1 and ϒA2 be two soft sets
as follows: over M. Then, the similarity measure of ϒA1 and ϒA2
1 + can be defined as by utilizing Euclidean distance,
n
dP (B1 , B2 ) = ( (|μB1 (i ) − μ+
B2 (i )|
2
2 1
i=1 Sim1 (ϒA1 , ϒA2 ) =
1 + d 3 (ϒA1 , ϒA2 )
+ |μ− − 1
B1 (i ) − μB2 (i )| ))
2 2
Another similarity measure can be defined as

Sim2 (ϒA1 , ϒA2 ) = e−σd


3 (ϒ ,ϒ )
A1 A2
Definition 3.2. [35] Let P1 and P2 be the set of param-
eters such that P1 =/ P2 , and (G1 , P1 ) and (G2 , P2 ) where σ is a positive real number called the steepness
be two fuzzy soft sets, then the similarity measure measure.
3174 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

Now we extend some existing distance measures to Theorem 3.6. Let D(A1 , A2 ) is the distance mea-
BFSSs in the following definition. H , d PH , d P NE
sure defined by dBFS BFS BFS and dBFS between
BFSS A1 and A2 , then D(A1 , A2 ) satisfy the
Definition 3.5. Let M = {1 , 2 , 3 , · · · n } be the following conditions:
universe, and P = {p1 , p2 , p3 , · · · pm } be the set of
parameters. Let A1 , A2 ⊆ P and A1 , A2 be two (i) (Nonnegativity) D(A1 , A2 ) ≥ 0 and
BFSSs over M with their BF-approximate function D(A1 , A2 ) = 0 ⇔ A1 = A2 ;
(ii) (Symmetry) D(A1 , A2 ) = D(A1 , A2 )
δA1 (pi ) = {(, μ+ −
A1 (), μA1 ()) :  ∈ M} (iii) (Triangle inequality) D(A1 , A2 ) ≤
D(A1 , A3 ) + D(A3 , A2 )
and
Proof. We only give proof for dBFS H . If  ,  ,
A1 A2
δA2 (pi ) = {(, μ+ −
A2 (), μA2 ()) :  ∈ M} A3 ∈ BFS(M), then

PY
H ( ,  ) ≥ 0
(i) dBFS A1 A2
respectively. Then, the distances of A1 and A2 are ∀i = 1, 2, · · · , m, j = 1, 2, · · · , n
defined as H ( ,  ) = 0
if dBFS A1 A2

1. Hamming distance: ⇒ |μ+ + −


A1 (pi )(j ) − μA2 (pi )(j )| + |μA1 (pi )(j ) −
μ−A2 (pi )(j )| = 0

=
H
dBFS (A1 , A2 )

1  +
2mn
m

i=1 j=1
n
(|μA1 (pi )(j ) − μ+
A2 (pi )(j )|
CO⇒ μ+
μ−
⇒ A1 = A2
Conversely, let
+ −
A1 (pi )(j ) = μA2 (pi )(j ) ∧ μA1 (pi )(j ) =
A2 (pi )(j )

 A1 = A2 ⇒ μ+ +
A1 (pi )(j ) = μA2 (pi )(j ) ∧
+ |μ− −
A1 (pi )(j ) − μA2 (pi )(i )|) − −
OR
μA1 (pi )(j ) = μA2 (pi )(j )
2. Poles-weighted Hamming distance: ⇒ |μ+ + −
A1 (pi )(j ) − μA2 (pi )(j )| + |μA1 (pi )(j ) −

PH
μA2 (pi )(j )| = 0
dBFS (A1 , A2 )
⇒ dBFSH ( ,  ) = 0
A1 A2
1  p +
m n
(ii) Clearly dBFS H ( ,  ) = d H ( ,  )
A1 A2 BFS A2 A1
= (κ |μA1 (pi )(j ) −μ+
A2 (pi )(j )|
TH

mn (iii) Triangular inequality follows easily from the


i=1 j=1
observation that for any three BFS-sets A1 , A2

+κ N
|μ−
A1 (pi )(j ) − μ−
A2 (pi )(j )|)
and A3 ,
∀i = 1, 2, · · · , m, j = 1, 2, · · · , n
|μ+ + −
A1 (pi )(j ) − μA2 (pi )(j )| + |μA1 (pi )(j ) −
AU

where κp + κN = 1. − + +
μA2 (pi )(j )| = |μA1 (pi )(j ) − μA3 (pi )(j ) +
3. Euclidean distance: μ+ + −
A3 (pi )(j ) − μA2 (pi )(j )| + |μA1 (pi )(j ) −
− − −
P
dBFS (A1 , A2 ) μA3 (pi )(j ) + μA3 (pi )(j ) − μA2 (pi )(j )|
≤ |μ+ + +
A1 (pi )(j ) − μA3 (pi )(j )| + |μA3 (pi )(j ) −
 1 
m 
n
+ − −
= [|μ+ + μA2 (pi )(j )| + |μA1 (pi )(j ) − μA3 (pi )(j )| +
A1 (pi )(j ) − μA2 (pi )(j )| ]
2
2m
i=1 j=1 |μ− −
A3 (pi )(j ) − μA2 (pi )(j )|
1 Therefore we have:
+ [|μ− −
A1 (pi )(j ) − μA2 (pi )(j )| ]
2 2 H ( ,  ) ≤ d H ( ,  ) +
dBFS A1 A2 BFS A1 A3
H ( ,  )
dBFS A3 A2
3. Normalized Euclidean distance: The others proof can be made similarly. 
NE
dBFS (A1 , A2 )
Theorem 3.7. Let BFS(M) be a set of all BFSSs over
 1 
m 
n H , d PH , d P
M. Then, each distance function dBFS BFS BFS
= [|μ+ +
A1 (pi )(j ) − μA2 (pi )(j )| ]
2
NE
and dBFS is a metric on BFS(M).
2mn
i=1 j=1

 21 Definition 3.8. Let A1 and A2 be two BFSSs


+ [|μ− −
A1 (pi )(j ) − μA2 (pi )(j )| ]
2
over M. Then, by using the Hamming distance,
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3175

similarity measures of A1 and A2 is defined as in {1, 2, · · · , n} and j ∈ {1, 2, · · · , m},
the following, then
SimBFS (A1 , A2 ) = 1
1
Sim (A1 , A2 ) =
1
.
1 + d H (A1 , A2 )
Example 3.10. Assume that M = {1 , 2 , 3 } is
Another similarity measure can be defined as the universe of discourse, P = {p1 , p2 , p3 , p4 } be
the set of parameters. Let A1 = {p1 , p2 , p3 } and
Sim2 (A1 , A2 ) = e−σd
H ( , )
A1 A2
A2 = {p1 , p2 , p3 } be subsets of P. Consider the
where σ is a positive real number called the steepness 
BFSSs defined over M

measure. A1 = p1 , {(1 , 0.110, −0.871),

(2 , 0.76, −0.500), (3 , 0.571, −0.211)} ,

PY
Definition 3.9. Let M = {1 , 2 , 3 , · · · , n } be
a universe, P = {p1 , p2 , p3 , · · · , pm } be the set p2 , {(1 , 0.811, −0.211),
  (2 , 0.790, −0.122),
of parameters, A1 , A2 ⊆ P and A1 , A2 be two (3 , 0.711, −0.002)} , p3 , {(1 , 0.920, −0.321),

BFSSs on M with their bipolar fuzzy approximate 
(2 , 0.530, −0.530), (3 , 0.560, −0.231)}
function

CO
and 
δA1 (pi ) = {(, μ+ −
A1 (), μA1 ()) :  ∈ M} 
A2 = p1 , {(1 , 0.240, −0.670),
and 
(
 2 , 0.333, −0.551), ( 3 , 0.521, −0.281)} ,
δA2 (pi ) = {(, μ+ −
A2 (), μA2 ()) :  ∈ M} p2 , {(1 , 0.870, −0.100),
  (2 , 0.571, −0.570),
(3 , 0.290, −0.221)} , p3 , {(1 , 0.821, −0.283),

respectively If A1 = A2 and μ+ A1 (pi )(j ) − 
OR
μ−A1 (p i )( j ) =/ 0 or μ +
A2 (p i )( j ) − μ −
A2 (pi )(j ) =
/ 0 (2 , 0.981, −0.022), (3 , 0.511, −0.421)} .
for at least one i ∈ {1, 2, · · · , n} and
j ∈ {1, 2, · · · , m}, The distances, by using Definition 3.5, of BFSSs A1
then the similarity measure between A1 and A2 is and A2 are calculated as follows:
defined by
TH

m n + − + −
i=1 j=1 |(μA1 (pi )(j ) + μA1 (pi )(j )).(μA2 (pi )(j ) + μA2 (pi )(j ))|
SimBFS (A1 , A2 ) = m n  + − + − 
j=1 max ||μA1 (pi )(j ) + μA1 (pi )(j )|| , ||μA2 (pi )(j ) + μA2 (pi )(j )||
2 2
i=1

where
AU

μ+
A1 (pi )(j )

= (μ+ + + H
(A1 , A2 ) = 0.47
A1 (pi )(1 ), μA1 (pi )(2 ), · · · , μA1 (pi )(n ))
dBFS

μ−
PH
dBFS (A1 , A2 ) = 0.32
A1 (pi )(j )

= (μ− − −
A1 (pi )(1 ), μA1 (pi )(2 ), · · · , μA1 (pi )(n )) E
dBFS (A1 , A2 ) = 0.381

μ+
A2 (pi )(j ) NE
dBFS (A1 , A2 ) = 0.213
= (μ+ + +
A2 (pi )(1 ), μA2 (pi )(2 ), · · · , μA2 (pi )(n )) and
1
μ− Sim1 (A1 , A2 ) = = 0.670
A2 (pi )(j ) 1 + d H ( A1 , A2 )

= (μ− − −
A2 (pi )(1 ), μA2 (pi )(2 ), · · · , μA2 (pi )(n )) and

If A1 = A2 and μ+ −
A1 (pi )(j ) + μA1 (pi )(j ) = 0
Sim2 (A1 , A2 ) = e−σd
H ( , )
+ −
and μA2 (pi )(j ) + μA2 (pi )(j ) = 0, for all i ∈
A1 A2
= 0.6866
3176 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

Now, we can obtain the following calculations: Theorem 3.12. Let P be the set of parameters,
μ+
A1 (p1 )(j ) = (0.110, 0.760, 0.571) A1 , A2 , A3 ⊆ P and A1 , A2 and A3 be three
μ−
A1 (p1 )(j ) = (−0.871, −0.500, −0.211)
BFSSs over M such that A1 is a BFS-subset of A2
μ+
A1 (p2 )(j ) = (0.811, 0.790, 0.711)
and A2 is BFS-subset of A3 . Then,
μ−
A1 (p2 )(j ) = (−0.211, −0.122, −0.002) SimBFS (A1 , A3 ) ≤ SimBFS (A2 , A3 )
μ+
A1 (p3 )(j ) = (0.920, 0.530, 0.560)
μ−
A1 (p3 )(j ) = (−0.321, −0.530, −0.231)
μ+
A2 (p1 )(j ) = (0.240, 0.333, 0.521) Proof. The proof is straightforward. 
μ−
A2 (p1 )(j ) = (−0.670, −0.551, −0.281)
μ+
A2 (p2 )(j ) = (0.870, 0.571, 0.290) Definition 3.13. Let M = {1 , 2 , 3 , · · · } be a uni-
μ− verse, P = {p1 , p2 , p3 , · · · } be the set of parameters,
A2 (p2 )(j ) = (−0.100, −0.570, −0.221)

PY
μ+ A1 , A2 ⊆ P and A1 , A2 be two BFSSs on M with
A2 (p3 )(j ) = (0.821, 0.981, 0.511)
their BF-approximate function
μ−
A2 (p3 )(j ) = (−0.283, −0.022, −0.421)
and
δA1 (pi ) = {(, μ+ −
A1 (), μA1 ()) :  ∈ M}
μ+ + μ− = (−0.761, 0.260, 0.360)

CO
A1 (p1 )(j ) A1 (p1 )(j )
μ+ + μ− = (0.600, 0.668, 0.709) and
A1 (p2 )(j ) A1 (p2 )(j )
μ+ + μ− = (−0.401, 0.000, −0.371)
A1 (p3 )(j ) A1 (p3 )(j ) δA2 (pi ) = {(, μ+ −
A2 (), μA2 ()) :  ∈ M}
μ+
A2 (p1 )(j ) + μ−
A2 (p1 )(j ) = (−0.430, −0.221, 0.240)
μ+
A2 (p2 )(j ) + μ−
A2 (p2 )(j ) = (0.770, 0.001, 0.069) respectively,
μ+
A2 (p3 )(j ) + μ−
A2 (p3 )(j ) = (0.538, 0.959, −0.110) If A1 = A2 , then the SM between A1 and A2 is
defined as
OR

m n + !
j=1 {μA1 (pi )(j ) . μ+ − −
A2 (pi )(j )} + {μA1 (pi )(j ) . μA2 (pi )(j )}
Sim+
i=1
BFS (A1 , A2 ) = m n + !
i=1 j=1 {μA1 (pi )(j ) ∨
2 μ+ 2 − 2 −
A2 (pi )(j ) } + {μA1 (pi )(j ) ∨ μA2 (pi )(j ) }
2

If A1 =/ A2 and P = A1 ∩ A2 = / φ, then we define


TH

Thus the similarity measure between A1 and A2 is δA1 (p) = 0 for p ∈ A2 \ P and δA2 (p) = 0 for p ∈
given by A1 \ P, then we compute Sim+ BFS (A1 , A2 ) by
SimBFS (A1 , A2 ) = 0.630 above formula.
Note: If P = A1 ∩ A2 = φ then Sim+ BFS (A1 ,
Remark: Theorems 3.11 and 3.12 hold for the simi- A2 ) = 0.
AU

larity measure given in Definition 3.9.


Example 3.14. We consider information given in
Theorem 3.11. Let P be the set of parameters, Example 3.10 and compute SM between A1 and
A1 , A2 ⊆ P and A1 and A2 be two BFSSs over A2 by the formula
M. Then, the following hold;

3 3 + !
j=1 {μA1 (pi )(j ) . μ+ − −
A2 (pi )(j )} + {μA1 (pi )(j ) . μA2 (pi )(j )}
Sim+
i=1
BFS (A1 , A2 ) = 3 3 + !
i=1 j=1 {μA1 (pi )(j ) ∨
2 μ+ 2 − 2 −
A2 (pi )(j ) } + {μA1 (pi )(j ) ∨ μA2 (pi )(j ) }
2

Here
μ+ +
A1 (p1 )(j ) . μA2 (p1 )(j ) = (0.110, 0.760, 0.571).
1. SimBFS (A1 , A2 ) = SimBFS (A1 , A2 )
(0.240, 0.330, 0.521) = 0.5747
2. 0 ≤ SimBFS (A1 , A2 ) ≤ 1
μ+ +
A1 (p2 )(j ) . μA2 (p2 )(j ) = (0.811, 0.790, 0.711).
3. SimBFS (A1 , A1 ) = 1
(0.870, 0.571, 0.290) = 1.362
μ+ +
A1 (p3 )(j ) . μA2 (p3 )(j ) = (0.220, 0.530, 0.560).
Proof. The proof is obvious by using Definition 3.9. (0.821, 0.981, 0.511) = 1.771
 μ− −
A1 (pi )(j ) . μA2 (pi )(j ) = (−0.871, −0.500,
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3177

− 0.211).(−0.670, −0.551, −0.281) = 0.9081 From here, there is σ ≥ 0 such that


μ− −
A1 (pi )(j ) . μA2 (pi )(j ) = (−0.211, −0.122,
− 0.002).(−0.100, −0.570, −0.281) = 0.0912
μ− −
A1 (pi )(j ) . μA2 (pi )(j ) = (−0.621, −0.530, [μ+ +
A1 (pi )(j ) . μA2 (pi )(j )] + σ
− 0.931).(−0.283, −0.022, −0.621) = 0.7655
Now the similarity between A1 and A2 is calcu- = [μ+ 2 +
A1 (pi )(j ) ∨ μA1 (pi )(j ) ].
2

lated as
Similarly, there exist ς ≥ 0 such that
Sim+
BFS (A1 , A2 ) = 0.6774
[μ− −
A1 (pi )(j ) . μA2 (pi )(j )] + ς
Remark: Theorems 3.15 and 3.16 hold for the simi-
larity measure given in Definition 3.13. = [μ− 2 −
A1 (pi )(j ) ∨ μA1 (pi )(j ) ].
2

PY
Theorem 3.15. Let P be the set of parameters, So that σ + ς = 0. This implies that σ = −ς which
A1 , A2 ⊆ P and A1 and A2 be two BFSSs over is not true. If, A1 = A2 , hence, it is proved that
M. Then, the followings hold; sim+BFS = 1
Converse can proved trivially from Definition 2. 
1. Sim+ +
BFS (A1 , A2 ) = SimBFS (A1 , A2 )
+

CO
2. 0 ≤ SimBFS (A1 , A2 ) ≤ 1 Theorem 3.16. Let P be the set of parameters,
3. Sim+
BFS (A1 , A2 ) = 1 ⇔ A1 = A2 A1 , A2 , A3 ⊆ P and A1 , A2 and A3 be three
BFSSs over M such that A1 is a BFS-subset of A2
Proof. Condition (1) and (2) can be proved trivially and A2 is BFS-subset of A3 . Then,
from the Definition 3.9. To prove (3) let us claim
that Sim+ Sim+ +
BFS (A1 , A3 ) ≤ SimBFS (A2 , A3 )
BFS (A1 , A2 ) = 1 when A1 = A2 .
OR
By considering the definition of simBFS the follow-
ing eqautions are obtained:

!
{μ+ + − −
A1 (pi )(j ) . μA2 (pi )(j )} + {μA1 (pi )(j ) . μA2 (pi )(j )}
! =1
{μ+ + − −
A1 (pi )(j ) ∨ μA1 (pi )(j ) } + {μA1 (pi )(j ) ∨ μA1 (pi )(j ) }
2 2 2 2
TH

!
⇒ {μ+ + − −
A1 (pi )(j ) . μA2 (pi )(j )} + {μA1 (pi )(j ) . μA2 (pi )(j )}

!
= {μ+ + − −
A1 (pi )(j ) ∨ μA1 (pi )(j ) } + {μA1 (pi )(j ) ∨ μA1 (pi )(j ) } .
2 2 2 2
AU

Now we claim the following equations

μ+ + + +
A1 (pi )(j ) . μA2 (pi )(j ) = μA1 (pi )(j ) ∨ μA1 (pi )(j )
2 2
Proof. Straightforward. 

and Definition 3.17. Let P be the set of parameters


and two BFSSs A1 and A2 are called γ-similar,
μ− − − −
A1 (pi )(j ) . μA2 (pi )(j ) = μA1 (pi )(j ) ∨ μA1 (pi )(j ) .
2 2
indicated as A1 tγ A2 , if SimBFS (A1 , A2 ) ≥ γ
where 0 < γ < 1.
Suppose Proposition 3.18. The relation of two BFSSs being
μ+ +
/ μ+
A1 (pi )(j ) . μA2 (pi )(j ) =
+
A1 (pi )(j ) ∨ μA1 (pi )(j )
2 2 γ-similar is reflexive and symmetric, but is not tran-
sitive.

exists because of the fact that Proof. Reflexive and symmetric properties follow
μ+ + + 2 +
A1 (pi )(j ) . μA2 (pi )(j ) ≤ μA1 (pi )(j ) ∨ μA1 from Definition 3 and given Example 3 shows that
(pi )(j )2 . tγ is not transitive. 
3178 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

Example 3.19. Let us suppose that M = {1 , 2 , 3 }


and P = {p1 , p2 , p3 }. Now, consider the BFS-
matrices given as below
⎛ ⎞
(0.270, −0.391) (0.421, −0.512) (0.611, −0.431)
⎜ ⎟
A1 = ⎝(0.251, −0.562) (0.580, −0.491) (0.920, −0.361)⎠ ,
(0.762, −0.233) (0.462, −0.480) (0.540, −0.212)
⎛ ⎞
(0.451, −0.212) (0.260, −0.891) (0.542, 0.390)
⎜ (0.290, −0.281) (0.460, −0.440) (0.641, −0.310)⎟
A2 =⎝ ⎠,
(0.272, 0. − 543) (0.280, −0.331) (0.890, −0.163)
⎛ ⎞

PY
(0.931, −0.150) (0.451, −0.590) (0.332, −0.144)
⎜(0.390, −0.282) (0.512, −0.551) (0.642, −0.272)⎟
A3 =⎝ ⎠.
(0.714, −0.323) (0.332, −0.184) (0.091, −0.560)

Take γ = 0.8415. Then, SimBFS (A2 , A1 ) =


0.9123 > 0.8415 and SimBFS (A1 , A3 ) = 0.8492
> 0.8415. But SimBFS (A2 , A3 ) = 0.8390
0.8415. This shows that γ-similar relation is not
transitive by using similarity measure given in
CO
If A1 = A2 , then weighted similarity measure
(WSM) between A1 and A2 is defined as
Definition 3.9.
WSMBFS (A1 , A2 ) = !
m n
OR
{μ+ + − −
A (pi )(j ) . μA (pi )(j )} + {μA (pi )(j ) . μA (pi )(j )}
n1 !.
i
× m i=1n j=1 1 2 1 2
j=1 i
i=1 j=1
{μ+ + 2 −
A (pi )(j ) ∨ μA (pi )(j ) } + {μA (pi )(j ) ∨
2 2 μ−
A2 (pi )(j ) }
2
1 2 1

Example 3.21. In Example 3.10, consider the weights


Definition 3.20. Let M = {1 , 2 , 3 , · · · } be a uni- 1 = 0.5, 2 = 0.1 and 3 = 0.4 for p1 , p2 and p3 ,
TH

verse, P = {p1 , p2 , p3 , · · · } be the set of parameters respectively. Then, the WSM between A1 and A2
and i be the weights of pi , A1 , A2 ⊆ P and is calculated as
A1 , A2 be two BFS- sets on M with their BFS-
approximate function defined as below, respectively: WSMBFS (A1 , A2 ) = 0.6785.
AU

Theorem 3.22. Let P be the set of parameters,


δA1 (pi ) = {(, μ+ −
A1 (), μA1 ()) :  ∈ M}
A1 , A2 ⊆ P and A1 and A2 be two BFSSs over
M. Then, the following relations hold;
and
1. WSMBFS (A1 , A2 ) = WSMBFS (A1 , A2 )
δA2 (pi ) = {(, μ+ −
A2 (), μA2 ()) :  ∈ M} 2. 0 ≤ WSMBFS (A1 , A2 ) ≤ 1
3. WSMBFS (A1 , A2 ) = 1 ⇔ A1 = A2 .

Proof. Proof can be done easily by using Definition 3.



Definition 3.23. Let M = {i : i = 1, · · · , m} and
P = {pj : j = 1, · · · , n}. If
⎛ ⎞
(μ+ −
11 , μ11 )1 (μ+ −
12 , μ12 )1 ··· (μ+ −
1n , μ1n )1
⎜ + − ⎟
⎜ (μ21 , μ21 )1 (μ+ −
22 , μ22 )1 ··· (μ+ −
2n , μ2n )1 ⎟
⎜ ⎟
1 = ⎜ .. .. .. .. ⎟
⎜ . . . . ⎟
⎝ ⎠
(μ+ −
m1 , μm1 )1 (μ+ −
m2 , μm2 )1 ··· (μ+ −
mn , μmn )1
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3179

and
⎛ ⎞
(μ+ −
11 , μ11 )2 (μ+ −
12 , μ12 )2 ··· (μ+ −
1n , μ1n )2
⎜ + − ⎟
⎜ (μ21 , μ21 )2 (μ+ −
22 , μ22 )2 ··· (μ+ −
2n , μ2n )2 ⎟
⎜ ⎟
2 = ⎜ .. .. .. .. ⎟
⎜ . . . . ⎟
⎝ ⎠
(μm1 , μ−
+
m1 )2 (μ+ −
m2 , μm2 )2 ··· (μ+ −
mn , μmn )2

are BFS-matrices of the BFSSs 1 = (1 , P) and


1 = (2 , P), then the proposed similarity measure
SimR between 1 and 2 is given as Proposition 3.25. The similarity measure defined in
Definition 3 satisfies the following:
< 1 , 2 >
SimR (1 , 2 ) = 1. 0 ≤ SimR (1 , 2 ) ≤ 1

PY
1 2 
2. SimR (1 , 2 ) = 1 ⇔ 1 = 2
where 3. SimR (1 , 2 ) = Sim(2 , 1 ).
< 1 , 2 > = i,j (μ+ − + −
ij , μij )1 .(μij , μij )2
Proof. We prove it for BFSSs (1 , P) and (2 , P)

CO
= tr(T1 2 ), defined over M and the set of parameters P
"
1  = < 1 , 1 >. with positive-membership and negative-membership
functions only.
Here tr(T1 2 ) is the sum of the elements in main Conditions 1 & 3 can be proved straightway from the
diagonal of the matrix T1 2 and is known as trace given Definition 3. Assume that
of the matrix T1 2 .
SimR (1 , 2 ) = 1.
OR

Example 3.24. Taking into consideration the Exam- This results in


ple 3.10, the corresponding BFS-matrices are written
as follows:
⎛ ⎞
(0.110, −0.871) (0.811, −0.211) (0.920, −0.321)
TH

⎜ ⎟
1 = ⎝(0.760, −0.500) (0.790, −0.122) (0.530, −0.530)⎠ ,
(0.571, −0.211) (0.711, −0.002) (0.560, −0.231)
⎛ ⎞
(0.240, −0.670) (0.870, −0.100) (0.821, −0.283)
⎜ (0.981, −0.022)⎟
2 = ⎝(0.330, −0.551) (0.571, −0.570)
AU

⎠.
(0.521, −0.281) (0.290, −0.221) (0.511, −0.421)

Now, we can deduce the following calculation:


< 1 , 2 > = (0.110, −0.871).(0.240, −0.670) + (0.811, −0.211).(0.870, −0.100) + · · ·
+(0.560, −0.231).(0.511, −0.421)
= 3.3853,
"
1  = (0.110)2 + (−0.871)2 + (0.811)2 + · · · + (−0.231)2
= 2.1943,
"
2  = (0.240)2 + (−0.670)2 + (0.870)2 + · · · + (−0.421)2
= 2.2093.
< 1 , 2 >
∴ SimR (1 , 2 ) =
1 2 
= 0.6983.
3180 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

< 1 , 2 > Corollary 3.29. The relation of two BFSSs being γ-


=1
1 2  similar is not an equivalence relation.

which yield (μ+ − + −


ij , μlm )1 = (μlm , μij )2 , for all val- Example 3.30. Suppose M = {1 , 2 , 3 } be a crisp
ues of l and m. Hence, it is clear that 1 = 2 . set and P = {p1 , p2 , p3 } be the set of parameters.
Conversely, suppose that 1 = 2 . Then, Consider the BFS-matrices

⎛ ⎞
(0.270, −0.391) (0.421, −0.512) (0.611, −0.431)
⎜(0.251, −0.562) (0.580, −0.491) (0.920, −0.361)⎟
(1 , P) = ⎝ ⎠,
(0.762, −0.233) (0.462, −0.480) (0.540, −0.212)
⎛ ⎞
(0.451, −0.212) (0.260, −0.891) (0.542, −0.390)

PY
⎜(0.290, −0.281) (0.460, −0.440) (0.641, −0.310)⎟
(2 , P) = ⎝ ⎠,
(0.272, −0.543) (0.280, −0.331) (0.890, −0.163)
⎛ ⎞
(0.931, −0.150) (0.451, −0.590) (0.332, −0.144)

CO
⎜(0.390, −0.282) (0.512, −0.551) (0.642, −0.272)⎟
(3 , P) = ⎝ ⎠.
(0.714, −0.323) (0.332, −0.184) (0.091, −0.560)

Take γ = 0.8515. Then, SimR (2 , 1 ) = 0.9012 >


0.8515 and SimR (1 , 3 ) = 0.8571 > 0.8515.
< 1 , 2 > = < 1 , 1 > But SimR (2 , 3 ) = 0.8417 0.8515. This shows
OR
= 1 2 that γ-similar relation is not transitive by using sim-
ilarity measure given in Definition 3.
= 1 2 .
< 1 , 2 >
∴ SimR (1 , 2 ) =
1 2  4. Weighted similarity measure for BFSSs
TH

= 1.
In the following section, we discuss weighted SM
 for two BFSSs and give some of its specific charac-
teristics.
Theorem 3.26. Suppose (1 , P), (2 , P) and
AU

(3 , P) be three BFSSs defined over M. If Definition 4.1. Let 1 and 2 be two BFSSs. Sup-
 2 , P) and (2 , P)⊆(
(1 , P)⊆(  3 , P), then pose that the weights of attributes pj are j ∈ [0, 1].
Sim (1 , 3 ) ≤ Sim (2 , 3 ).
R R
The weighted similarity measure SimR W between 1
and 2 is defined by
Proof. Straightforward. 
< 1 , 2 >
W (1 , 2 ) =
SimR
Definition 3.27. Two BFSSs (1 , P1 ) and (2 , P2 ) 1 2 
defined over (M, P) are called γ-similar, indicated as
(1 , P1 ) tγ (2 , P2 ), if SimR (1 , 2 ) ≥ γ where where
0 < γ < 1. + − + −
i,j j (μij , μij )1 .(μij , μij )2
< 1 , 2 > = ,
Theorem 3.28. The relation of two BFSSs being j j
γ-similar is reflexive and symmetric, but is not tran- "
sitive. 1  = < 1 , 1 >.

Proof. Proposition 3 and Theorem 3 proves the


symmetry and reflexivity properties of BPSSs and Example 4.2. Use BFSSs described in Example 3,
Example 3 justifies the non-transitivity of BPFSSs. and take 1 = 0.22, 2 = 0.51 and 3 = 0.42 as
 weights of attributes p1 , p2 and p3 . We obtain,
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3181

< 1 , 2 > = 2.0672, 9. A3HY (1 , 2 ) = 


1  = 1.9020, 1−max |μ+ + − −
 (i )−μ (i )|,|μ (i )−μ (i )|
 1 2 1 2 
1+max |μ+ + − −
1 (i )−μ2 (i )|,|μ1 (i )−μ2 (i )|
2  = 1.5479,
10. S(1 , 2 ) =
∴ SimR i [μ+ + − −
W (1 , 2 ) = 0.6962.  (i )μ (i )+μ (i )μ (i )]
1 2 1 2
i [{μ+ + − −
 (i ) ∨μ (i ) }+{μ (i ) ∨μ (i ) }]
2 2 2 2
1 2 1 2

Proposition 4.3. The weighted similarity measure


proposed in Definition 4 holds the following condi- Example 4.4. Take M = {1 , 2 , 3 } and P =
tions: {p1 , p2 , p3 } as the universe and aggregate of prop-
erties, respectively. Take a look at the BFSSs that
1. 0 ≤ SimR
W (1 , 2 ) ≤ 1 have matrix representations as follow:.
W (1 , 2 ) = 1 ⇔ 1 = 2
2. SimR

PY
(1 , P)
W (1 , 2 ) = SimW (2 , 1 ).
3. SimR ⎛ ⎞
(0.62, −0.21) (0.42, −0.52) (0.81, −0.12)
⎜ (0.64, 0.34) ⎟
= ⎝(0.51, −0.32) (0.71, −0.11) ⎠,
Proof. The proof is obvious by Definition 4. 
(0.83, −0.42) (0.63, 0.0) (0.92, 0.0)

4.1. Comparison analysis

Muthukumar and Krishnan [36] developed some


similarity measures for intuitionistic fuzzy soft sets.
CO (2 , P)


(0.54, −0.32)
= ⎝(0.62, −0.22)
(0.92, 0.0)
(0.71, 0.0)
(0.42, 0.0)
(0.54, −0.11)
(0.62, 0.31)
(0.56, −0.11)⎟
(0.81, 0.0)
⎠,

Now we extend these similarity measures towards


bipolar fuzzy soft sets with some improvements in the (3 , P)
OR
existing similarity measures. The extended similarity ⎛ ⎞
(0.44, −0.42) (0.61, −0.22) (0.54, −0.11)
measures for bipolar fuzzy soft sets are the following:
⎜ (0.54, 0.41) ⎟
= ⎝(0.32, −0.32) (0.74, 0.11) ⎠.
1. AC (1 , 2 ) = 1 − 2n i=1 |(μ+
1 n
1 (i ) − (0.21, 0.0) (0.52, 0.0) (0.11, −0.82)
+ −
μ2 (i )) − (μ1 (i ) − μ− (
2 i ))|
2. AH (1 , 2 ) = 1 − 2n i=1 (|μ+
1 n
1 (i ) −
TH

The Figure 1 and Table 2 shows that the proposed SM


μ+ (
2 i )| + |μ −
(
1 i ) − μ −
(
2 i )|) SimR given in Definition 3.23 provides a significant
3. AL (1 , 2 ) = 1 − 4n i=1 |(μ+
1 n
1 (i ) −
rating as compared to existing SMs.
μ+ (
2 i )) + (μ −
(
1 i ) − μ −
(
2 i ))| −
+ + −
4n i=1 (|μ1 (i ) − μ2 (i )| + |μ1 (i ) −
1 n
AU


μ2 (i )|)
#O (1 , 2 ) = 1 −
4. A
+ + − −
2n i=1 {(μ1 (i ) − μ2 (i )) + (μ1 (i ) − μ2 (i )) }
1 n 2 2

$ 1 , 2 ) %=
5. ADC (
%p
+ +
p 1 n % μ1 (i )+1−μ2 (i ) μ− (i )+1−μ−
2 (i ) %
1− n i=1 % 2 − 1 2 %
6. AHB (1 , 2 ) =
⎧ "   " 
⎨ 1− p n1 ni=1 |μ+ (i )−μ+ (i )|p 1
+ 1− p − −
n i=1 |μ (i )−μ (i )|
n p
1 2 1 2
, if p =
/ 1
⎩ 2
AH (1 , 2 ), if p = 1

7. A1HY (1 , 2 ) = 1 − n1 ni=1 max |μ+ 1 (i ) −
+ − − 
μ2 (i )|, |μ1 (i ) − μ2 (i )|
8. A2HY(1 , 2 ) =
 + + − −

exp − n1 ni=1 max |μ (i )−μ (i )|,|μ (i )−μ (i )| −exp(−1)
1 2 1 2
1−exp(−1)
3182 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

Table 2 rise in competition. The specificity of the pharma-


Comparative analysis of proposed SM with existing SMs ceutical industry is a sector that requires high-level
(1 , 2 ) (2 , 3 ) (1 , 3 ) transport and logistics facilities, such as the protec-
AC 0.5657 0.3933 0.3600 tion and continuous observation of pharmaceutical
AH 0.5433 0.3467 0.300 products due to its important relationship with health
AL 0.5600 0.3700 0.3200
AO 0.6783 0.4383 0.4031 and life, especially during the COVID-19 pandemic,
ADC 0.7072 0.509 0.507 it is extremely relevant. It is more critical than any-
AHB 0.6793 0.4406 0.441 thing tracking down all the agents from production
1
SHY 0.4233 0.0000 0.03233
2 to patients with utmost care since health care qual-
SHY 0.3356 0.0000 0.0208
3
SHY 0.2666 0.0000 0.0170 ity and pharmaceuticals are essential and lives are at
S 0.8229 0.5007 0.4943 stake. Products to patients or customers from the pro-
SimR 0.9420 0.7237 0.73172 duction business. In the pharmaceutical supply chain,

PY
there are many stages. The raw materials are sent to
the pharmaceutical plant, where the drugs are pro-
duced and packaged before being sold. Following
that, a wide shipment is carried out to supply patient
medications to local warehouses. So, in the case of

CO
logistics of pharmaceutical care, there is a great deal
of meaning in human life.
There are many problems that pharmaceutical
logistics must face, such as supply chain visibility.
The big safety issue is the theft and misplacement
of medication. The pharmaceutical industry will face
challenges with maintaining its integrity if the com-
OR
pany is unable to abolish such supply chain leaks. To
Fig. 1. Comparison Chart.
comply with the criteria of the High Quality Standard
(HQS), the logistics service provider must be consis-
5. Application to pharmaceutical logistics and tent and reliable. This attribution is more critical than
supply chain management reliability in the logistics service provider selection
TH

process (LSP).
Case Study: The major part of the overall market plan of the
The COVID-19 pandemic has certainly created both organization is the distribution of medication. LSP
critical health and economic crises. Its impact on should be safe and reliable in order to show quality
logistics companies is much significant. During these and reliability of products and services. Besides, to
AU

tremendous times, pharmaceutical logistics faced ensure regularity in the pharmaceutical supply chain
rough challenges as the nature of the industry is also helps productivity of the pharmaceutical indus-
very spontaneous sales slowed down due to national try. In the delivery of medication in the area, phase is
closures in the affected areas. The supply of mate- critical. The timely supply of pharmaceutical prod-
rials and stocks, as well as the import and export ucts helps to satisfy the increasing demand and to
of pharmaceutical, have been severely affected. Due avoid the lack of basic medicine requirements.
to inactive manufacturing activities, delays in the Innovation is an instrument that improves per-
delivery of goods have a negative impact on major formance, lowers costs and creates competitive
management operations. To address all these uncer- advantages. As a result, the pharmaceutical indus-
tainties in the logistics aspects, company strategy try is looking for the best LSP to pay more attention
should be updated with the demand of the customers to changes. The key problem is that LSP would not
and always try to thrive while identifying their char- be apt to maintain solid, profitable and long-lasting
acteristics features with care. partnerships without sustaining innovation. The regu-
The main aim of each logistics company is to lation of temperature is also a major factor that differs
achieve a strong partnership between business part- supply chains for pharmaceutical products from other
ners and customers, exceeding their expectations or supply chain networks.
even surpassing the standard requirements of the cus- Keeping the temperature optimized to ensure the
tomers in the aiding process as a result of a dramatic quality and standard of pharmaceutical products
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3183

during the transport and warehousing process is


mandatory.
Because of the rapid increase in sales volumes of
pharmaceutical products, it is not necessary to extend
room for warehousing. Although logistics companies
are not in the current scenario, it is important to meet
the rise in demands of the pharmaceutical industry. In
order to meet the requirements of increased visibility
and overall cost savings, they do not have appro- Fig. 2. Flow chart of Algorithm 1.
priate infrastructure. Huge investments are required
for such specialized warehouses with temperature-
controlled zones. With a more developed set-up, big best logistic company that meets their certain crite-
ria among five logistic companies L1 , L2 , L3 , L4 , L5

PY
corporations already exist in this market, making
it extremely difficult for new businesses to enter. with a set of criteria C = {C1 , C2 , C3 } as follows:
Besides, reliability and confidentiality of data, knowl- C1 = Consistency and reliability
edge, and expertise shared between manufacturers
and firms in logistics is essential. C2 = Timely supply & good distribution practices

CO
Algorithm 1 C3 = Innovation, online tracking system.

Step 1: Consider the set of logistic companies Consider the set of parameters
L = {L1 , L2 , · · · , Ln }, the set of param-
P = {p1 = unsatisfied, p2 = satisfied,
eters P = {p1 , p2 , · · · , pm }, the set of
criteria C = {C1 , C2 , · · · , Ct }, and set of p3 = highly satisfied}
decision-maker D = {D1 , D2 , · · · , Dk },
OR
where t, m, k are positive integers. which is linguistic labels used to describe satisfac-
Step 2: Write a standard BFSS (, P) as a model tion level of decision-making in selecting logistic
BFSS corresponding the requirements of company.
the pharmaceutical company. Assume that (, P) is the standard BFSS treated as
Step 3: Write the BFS-decision matrices corre- model BFSS corresponding the requirements of the
TH

sponding to the assessment of DMs in pharmaceutical company. If the logistic company is


terms of BFSSs (i , P) corresponding to more efficient than the expectation level of customer,
the profile of each logistic company. then positive grade is given as large as possible. Like-
Step 4: Write the unanimous decision matrix in wise, if the logistic company is less efficient than the
terms of (i , P) and standard BFSS expectation level of customer, then a negative grade
AU

(, P). is assigned as low as possible. To select the best


Step 5: Find SM for each BFSS (i , P) with the logistic company to transport pharmaceutical prod-
model BFSS (, P) using Definition 3.23. ucts, the company consults two decision-makers Dk ,
Step 6: Select the threshold value γ ∈]0, 1[. (k = 1, 2). Assume that (i , P) be the BFSS profile
Step 7: The logistic company with similarity mea- of the logistic companies Li , (i = 1, 2, · · ·, 5) with
sure ≥ γ is the best company to be choose. parameters pm , (m = 1, 2, 3) under the criteria Ct ,
Step 8: Write the final ranking and optimal deci- (t = 1, 2, 3). Then, we find unanimous BFS decision
sion. set according to the necessary criterion. To get the
SM between these BFSSs, we will match the charac-
teristics of the logistic companies with the model in
Table 5.
Figure 2 shows the procedural steps of Algorithm 1. Table 3 and Table 4 gives BFS decision matri-
ces. Let γ = 0.7 be the threshold value. Compare
5.1. Numerical example: this value with the calculated value of SM in Table
5. Then, we see that the model BFSS (, P) is sig-
A pharmaceutical company wants to select a nificantly similar to (2 , P) and (5 , P). The other
logistic company for transportation of its products. BPFSS are not similar to it. We note that the logistic
Company consults two decision-makers to choose the companies L2 and L5 are the best logistic companies
3184
Table 3
BFS decision matrix 1

M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application
(p1 , C1 ) (p1 , C2 ) (p1 , C3 ) (p2 , C1 ) (p2 , C2 ) (p2 , C3 ) (p3 , C1 ) (p3 , C2 ) (p3 , C3 )
(1 , P) (0.381, -0.845) (0.380, -0.842) (0.542, -0.681) (0.570, -0.391) (0.681, -0.271) (0.740, -0.512) (0.311, -0.521) (0.911, -0.382) (0.890, -0.415)
(2 , P) (0.542, -0.291) (0.121, -0.760) (0.482, -0.719) (0.823, -0.291) (0.562, -0.551) (0.342, -0.634) (0.861, -0.252) (0.991, -0.132) (0.814, -0.531)
(3 , P) (0.425, -0.261) (0.152, -0.391) (0.331, -0.824) (0.858, -0.046) (0.091, -0.331) (0.382, -0.425) (0.132, -0.942) (0.258, -0.910) (0.110, -0.970)
(4 , P) (0.212, -0.512) (0.526, -0.367) (0.632, -0.110) (0.730, -0.411) (0.290, -0.281) (0.536, -0.431) (0.551, -0.821) (0.821, -0.552) (0.501, -0.492)
(5 , P) (0.542, -0.210) (0.333, -0.490) (0.121, -0.621) (0.932, -0.152) (0.661, -0.453) (0.122, -0.302) (0.133, -0.561) (0.160, 0.931) (0.262, -0.480)

AU
TH
OR
CO
Table 4
BFS decision matrix 2
PY
(p1 , C1 ) (p1 , C2 ) (p1 , C3 ) (p2 , C1 ) (p2 , C2 ) (p2 , C3 ) (p3 , C1 ) (p3 , C2 ) (p3 , C3 )
(1 , P) (0.322, -0.321) (0.433, -0.352) (0.429, -0.071) (0.228, -0.451) (0.115, -0.518) (0.471, -0.228) (0.588, -0.129) (0.661, -0.428) (0.317, -0.191)
(2 , P) (0.117, -0.225) (0.342, -0.521) (0.192, -0.810) (0.261, -0.517) (0.451, -0.284) (0.622, -0.344) (0.251, -0.177) (0.153, -0.462) (0.163, -0.544)
(3 , P) (0.295, -0.610) (0.175, -0.721) (0.313, -0.531) (0.426, -0.522) (0.627, -0.215) (0.221, -0.613) (0.214, -0.519) (0.814, -0.013) (0.625, -0.352)
(4 , P) (0.113, -0.024) (0.641, -0.643) (0.112, -0.359) (0.329, -0.525) (0.454, -0.513) (0.512, -0.054) (0.880, -0.036) (0.172, -0.604) (0.351, 0.652)
(5 , P) (0.100, -0.515) (0.261, -0.356) (0.820, -0.544) (0.414, -0.152) (0.803, -0.251) (0.190, -0.470) (0.251, -0.809) (0.103, -0.055) (0.941, -0.281)
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3185

Table 6

(0.263, -0.534)
(0.890, -0.415)
(0.163, -0.544)
(0.625, -0.352)

(0.941, -0.281)
(0.351, 0.652)
SM between BFSS model and
pharmaceutical logistics

(p3 , C3 )
Logistic companies SimR (, i )
L1 0.6092
L2 0.8108
(0.253, -0.462) L3 0.6678
(0.991, -0.428)
(0.153, -0.462)
(0.814, -0.013)
(0.172, -0.604)
(0.103, -0.055)
L4 0.6903
L5 0.7993
(p3 , C2 )

Table 7
Weighted SM between BFSS model
(0.321, -0.207)
(0.588, -0.521)
(0.251, -0.177)
(0.214, -0.519)
(0.880, -0.036)
(0.251, -0.809)

PY
and pharmaceutical logistics
(p3 , C1 )

Logistic companies SimR


W (, i )
L1 0.6292
L2 0.8308
L3 0.6578
L4 0.6913

CO
(0.642, -0.354)
(0.740, -0.512)
(0.622, -0.344)
(0.221, -0.613)
(0.512, -0.054)
(0.190, -0.470)

L5 0.7893
(p2 , C3 )

that meet the certain criteria. Since L2 is most similar


Unanimous BFS decision matrix

(0.500, -0.294)
(0.681, -0.271)
(0.451, -0.284)
(0.627, -0.215)
(0.454, -0.513)
(0.803, -0.251)

to model in that case pharmaceutical company should


OR
choose it as logistic service provider for their goods.
(p2 , C2 )

The SMs between BFS-model and logistic compa-


Table 5

nies are given in Table 6.


The ranking of pharmaceutical logistic companies
in view of similarity by Table 6 becomes
(0.320, -0.547)
(0.570, -0.451)
(0.261, -0.517)
(0.426, -0.522)
(0.329, -0.525)
(0.414, -0.152)

TH
(p2 , C1 )

L2  L5  L4  L3  L1 .

Weighted SM between BFSS model and pharma-


ceutical logistics is expressed by Table 7. Now, we
(0.510, -0.759)
(0.542, -0.681)
(0.482, -0.719)
(0.313, -0.531)
(0.112, -0.359)
(0.820, -0.544)

AU

consider the weights w1 = 0.321, w2 = 0.524 and


(p1 , C3 )

w3 = 0.812 corresponding to p1 , p2 and p3 , respec-


tively, to compute the weighted SMs. Table 6 presents
the weighted similarity measure of each pharmaceu-
tical logistic company. We see that the ranking of
(0.364, -0.800)
(0.433, -0.842)
(0.342, -0.760)
(0.175, -0.721)
(0.641, -0.643)
(0.261, -0.356)

logistic companies by using weighted similarity mea-


(p1 , C2 )

sure becomes as follows:

L2  L5  L4  L3  L1 .
(0.532, -0.251)
(0.381, -0.845)
(0.542, -0.291)
(0.295, -0.610)
(0.113, -0.024)
(0.100, -0.515)

Here, Figure 3 shows the rankings of these two mea-


(p1 , C1 )

sures for each pharmaceutical logistic company.


These two rankings of logistic companies show that
L2 is more efficient. We note that the ranking of logis-
tics with high preference becomes L5 , L4 , L3 , and L1
(2 , P)
(1 , P)

(3 , P)
(4 , P)
(5 , P)
(, P)

which is similar by both suggested SMs, i.e. SimR and


SimR W.
3186 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

[5] W.R. Zhang, Bipolar fuzzy sets and relations: a computa-


tional framework for cognitive modeling and multiagent
decision analysis, NAFIPS/IFIS/NASA94. Proceedings of
the First International Joint Conference of The North
American Fuzzy Information Processing Society Biannual
Conference. The Industrial Fuzzy Control and Intellige
(1994), 305–309.
[6] W.R. Zhang, Bipolar fuzzy sets, Proceedings of IEEE Inter-
national Conference on Fuzzy Systems (1998), 835–840.
[7] R.R. Yager, Pythagorean fuzzy subsets, IFSA World
Congress and NAFIPS Annual Meeting (IFSA/NAFIPS),
2013 Joint, Edmonton, Canada, IEEE, (2013), 57–61.
[8] R.R. Yager and A.M. Abbasov, Pythagorean membership
Fig. 3. Bar chart of SimR and SimR
W of pharmaceutical logistics. grades, complex numbers, and decision making, Interna-
tional Journal of Intelligent Systems 28(5) (2013), 436–452.

PY
[9] R.R. Yager, Pythagorean membership grades in multi-
criteria decision making, IEEE Transactions on Fuzzy
Systems 22(4) (2014), 958–965.
6. Conclusion [10] R.R. Yager, Generalized orthopair fuzzy sets, IEEE Trans-
actions on Fuzzy Systems 25(5) (2017), 1220–1230.
The distance and similarity measure for bipolar [11] F. Smarandache, Neutrosophy: Neutrosophic Probability,

CO
fuzzy soft sets (BFSSs) are developed to resolve Set, and Logic: Analytic Synthesis & Synthetic Analysis,
American Research Press, (1998).
various uncertain real-life problems. For these objec- [12] F. Smarandache, Extension of soft set to hypersoft set, and
tives, we developed a new distance measure and then to plithogenic hypersoft set, Neutrosophic Sets and Sys-
similarity measures (SMs) with the help of cosine tems 22 (2018), 168–170. DOI: 10.5281/zenodo.2159754
[13] T. Mahmood, K. Ullah, Q. Khan and N. Jan, An Approach
similarity, inner product of matrices, and weighted
towards decision making and medical diagnosis problems
measures for bipolar fuzzy soft sets (BFSSs). Some using the concept of spherical fuzzy sets, Neural Computing
essential features of the proposed measures are inves-
OR
and Applications 31 (2019), 7041–7053.
tigated, including the measure of two similar BFSSs [14] S. Ashraf, S. Abdullah, T. Mahmood, F. Ghani and T.
Mahmood, Spherical fuzzy sets and their applications in
equal to the unity. A relationship of tγ between two multi-attribute decision making problems, Journal of Intel-
BFFs is used in this SM. Additionally, an advanced ligent & Fuzzy Systems 36(3) (2019), 2829–2844.
multi-attribute decision-making (MADM) algorithm [15] M. Rafiq, S. Ashraf, S. Abdullah, T.Mahmood and M.
is designed for pharmaceutical logistics and supply Shakoor, The cosine similarity measures of spherical fuzzy
TH

sets and their applications in decision making, Journal of


chain management in COVID-19. Furthermore, the Intelligent & Fuzzy Systems 36(6) (2019), 6059–6073.
validity of the proposed MADM method is estab- [16] F.K. Gundogdu and C. Kahraman, Properties and arithmetic
lished for the selection of the best pharmaceutical operations of spherical fuzzy sets, Decision Making with
logistics company, and a comparative analysis of Spherical Fuzzy Sets: Theory and Applications, Studies in
Fuzziness and Soft Computing (2021), 3–25.
the proposed similarity measures with some of the [17] G. Wei, C. Wei and H. Gao, Multiple attribute decision
AU

existing similarity measures is demonstrated. Apart making with interval-valued bipolar fuzzy information and
from the implementation viewpoint, the proposed their application to emerging technology commercialization
measures have considerable potential for further the- evaluation, IEEE Access 6 (2018), 60930–60955.
[18] Y. Han, Z. Lu, Z. Du, Z. Luo and S. Chen, A Yin Yang bipolar
oretical exploration. The proposed measures could be fuzzy cognitive TOPSIS method to bipolar disorder diagno-
effectively implemented with minor modifications to sis, Computer Methods and Programs in Biomedicine 158
other extensions of fuzzy sets. (2018), 1–10.
[19] S. Abdullah, M. Aslam and K. Ullah, Bipolar fuzzy soft sets
and its applications in decision making problem, Journal of
Intelligent & Fuzzy Systems 27(2) (2014), 729–742.
[20] F. Karaaslan and S. Karatas, A new approach to bipolar soft
References
sets and its applications, Discrete Mathematics, Algorithms
and Applications 07(4) (2015), 1550054.
[1] L.A. Zadeh, Fuzzy sets, Inform and Control 8 (1965), [21] M. Naz and M. Shabir, On fuzzy bipolar soft sets, their
338–356. algebraic structures and applications, Journal of Intelligent
[2] K.T. Atanassov, Intuitionistic fuzzy sets, Fuzzy Sets and & Fuzzy Systems 26(4) (2014), 1645–1656.
Systems 20 (1986), 87–96. [22] M. Akram and N. Waseem, Similarity measures for new
[3] K.T. Atanassov, Intuitionistic fuzzy sets: Theory and appli- hybrid models: mF Sets and mF Soft Sets, Punjab University
cation, Studies in Fuzziness and Soft Computing, XVIII, Journal of Mathematics 51(6) (2019), 115–130.
Physica Verlag, Heidelberg, (1999), 324–330. [23] K.M. Lee, Bipolar-valued fuzzy sets and their basic oper-
[4] D. Molodtsov, Soft set theory-first results, Comput. Math. ations, Proceedings of International Conference, Bangkok,
Appl. 37(4-5) (1999), 19–31. Thailand, (2000), 307–317.
M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application 3187

[24] K.M. Lee, Comparison of interval-valued fuzzy sets, intu- [45] Y. Jun, Cosine similarity measures for intuitionistic fuzzy
itionistic fuzzy sets, and bipolar-valued fuzzy sets, Journal sets and their applications, Mathematical and Computer
of Korean Institute of Intelligent Systems 14(2) (2004), Modeling 53(1-2) (2011), 91–97.
125–129. [46] Z. Liang and P. Shi, Similarity measures on intuitionistic
[25] I. Batyrshin, Towards a general theory of similarity and fuzzy sets, Pattern Recognit. Lett. 24 (2003), 2687–2693.
association measures: Similarity, dissimilarity and correla- [47] W.L. Hung and M.S. Yang, Similarity measures of intuition-
tion functions, Journal of Intelligent & Fuzzy Systems 36(4) istic fuzzy sets based on Lp metric, Int. J. Approx. Reason.
(2019), 2977–3004. 46 (2007), 120–136.
[26] I.Z. Batyrshin, Data ccience: Similarity, dissimilarity and [48] Z.S. Xu and X. Cai, Intuitionistic fuzzy information aggre-
correlation functions, In Artificial Intelligence. Lecture gation: theory and applications, Springer Heidelberg New
Notes in Computer Science, Springer, Cham, 11866 (2019), York Dordrecht London, (2012).
13–28. [49] J. Wang, H. Gao and G. Wei, The generalized dice sim-
[27] M. Akram, Bipolar fuzzy graphs, Information Sciences ilarity measures for Pythagorean fuzzy multiple attribute
181(24) (2011), 5548–5564. group decision making, International Journal of Intelligent
[28] M.A. Alghamdi, N.O. Alshehri and M. Akram, Multi- Systems 34(6) (2019), 1158–1183.

PY
criteria decision-making methods in bipolar fuzzy environ- [50] Z. Hussain and M.S. Yang, Distance and similarity measures
ment, Int. J. Fuzzy Syst. 20(6) (2018), 20–57. of Pythagorean fuzzy sets based on the Hausdorff metric
[29] M. Riaz and S.T. Tehrim, On bipolar fuzzy soft topol- with application to fuzzy TOPSIS, International Journal of
ogy with decision-making, Soft Computing 24(24) (2020), Intelligent Systems 34(10) (2019), 1–22.
18259–18272. [51] Q. Zhang, J. Hu, J. Feng, A. Liu and Y. Li, New similarity
[30] M. Riaz and S.T. Tehrim, Bipolar fuzzy soft mappings with measures of Pythagorean fuzzy sets and their applications,
application to bipolar disorders, International Journal of IEEE Access 7 (2019), 138192–138202.

CO
Biomathematics 12(7) (2019), 1–31. [52] E.P. Augustine, New similarity measures for Pythagorean
[31] J.C.R. Alcantud, G. Santos-García, X.D. Peng and J. Zhan, fuzzy sets with applications, International Journal of Fuzzy
Dual extended hesitant fuzzy sets, Symmetry 11(5) (2019), Computation and Modelling 3(1) (2020), 75–94.
1–13. [53] X.T. Nguyen, V.D. Nguyen, V.H. Nguyen and H. Garg,
[32] X.L. Zhang and Z.S. Xu, Extension of TOPSIS to multi- Exponential similarity measures for Pythagorean fuzzy sets
ple criteria decision making with Pythagorean fuzzy sets, and their applications to pattern recognition and decision-
International Journal of Intelligent Systems 29 (2014), making process, Complex & Intelligent Systems 5 (2019),
1061–1078. 217–228. https://doi.org/10.1007/s40747-019-0105-4.
OR
[33] A. Kharal, Distance and similarity measures for soft sets, [54] M. Riaz, K. Naeem and D. Afzal, A similarity measure
New Math. Nat. Comput. (2010), 1–14. under Pythagorean fuzzy soft environment with applica-
[34] H. Kamaci, Similarity measure for soft matrices and its tions, Computational and Applied Mathematics 39 (2020),
applications, Journal of Intelligent & Fuzzy Systems 36(4) 1–17.
(2019), 3061–3072. [55] N. Jan, L. Zedam, T. Mahmood, E. Rak and Z. Ali, General-
[35] P. Majumdar and S.K. Samanta, Similarity measure of soft ized dice similarity measures for q-rung orthopair fuzzy sets
sets, New Math. Nat.Comput. 4(1) (2008), 1–12. with applications, Complex & Intelligent Systems 6 (2020),
TH

[36] P. Muthukumar and G.S.S. Krishnan, A similarity measure 545–558.


of intuitionistic fuzzy soft sets and its application in medical [56] H. Garg, Z. Ali and T. Mahmood, Generalized dice similar-
diagnosis, Applied Soft Computing 41 (2016), 148–156. ity measures for complex q-Rung Orthopair fuzzy sets and
[37] M. Saqlain, M. Riaz, M.A. Saleem, M.S. Yang, Distance and its application, International Journal of Intelligent Systems,
similarity measures for neu-trosophic hypersoft set (NHSS) (2020), https://doi.org/10.1007/s40747-020-00203-x.
with construction of NHSS-TOPSIS and applications, IEEE [57] P. Wang, J. Wang, G. Wei and C. Wei, Similarity measures
AU

Access 9 (2021), 30803–30816. of q-rung orthopair fuzzy sets based on cosine func-
[38] S.M. Chen, Measures of similarity between vague sets, tion and their applications, Mathematics 7(4) (2019), 340;
Fuzzy Sets Syst. 74 (1995), 217–223. https://doi.org/10.3390/math7040340.
[39] S.M. Chen, Similarity measures between vague sets and [58] D. Liu, X. Chen and D. Peng, Some cosine similarity mea-
between elements, IEEE Trans. Syst. Man Cybern. B 27(1) sures and distance measures between q-rung orthopair fuzzy
(1997), 153–168. sets, Mathematics 34(7) (2019), 1572–1587.
[40] S.M. Chen, M.S. Yeh and P.Y. Hsiao, A comparison of sim- [59] V. Patrascu, Similarity, cardinality and entropy for bipolar
ilarity measures of fuzzy values, Fuzzy Sets and Systems 72 fuzzy set in the framework of pentavalued representation,
(1995), 79–89. arxiv, (2015).
[41] L.K. Hyung, Y.S. Song and K.M. Lee, Similarity measure [60] V. Ulucay, I. Deli and M. Sahin, Similarity measures of
between fuzzy sets and between elements, Fuzzy Sets and bipolar neutrosophic sets and their application to multiple
Systems 62(3) (1994), 291–293. criteria decision making, Neural Computing & Applications
[42] D. Li and C. Cheng, New similarity measures on intuitionis- 29 (2018), 739–748.
tic fuzzy sets and application to pattern recognition, Pattern [61] R.M. Hashim, M. Gulistan and F. Smarandache, Applica-
Recognit. Lett. 23 (2002), 221–225. tions of neutrosophic bipolar fuzzy sets in hope foundation
[43] Y.H. Li, D.L. Olson and Q. Zheng, Similarity measures for planning to build a children hospital with different types
between intuitionistic fuzzy (vague) sets: a compara- of similarity measures, Symmetry 10(8) (2018), 1–26.
tive analysis, Pattern Recognition Letters 28(2) (2007), [62] S. Petchimuthu and H. Kamaci, The row-products of inverse
278–285. soft matrices in multicriteria decision making, Journal of
[44] D.H. Hong and C. Kim, Note on similarity measure between Intelligent & Fuzzy Systems 36(6) (2019), 6425–6441.
vague sets and elements, Information Sciences 115(1–4) [63] H. Kamaci, Introduction to N-soft algebraic structures,
(1999), 83–96. Turkish Journal of Mathematics 44(6) (2020), 2356–2379.
3188 M. Riaz et al. / Distance and similarity measures for bipolar fuzzy soft sets with application

[64] H. Kamaci and S. Petchimuthu, Bipolar N-soft set the- [67] Z. Zararsiz, Similarity measures of sequence of fuzzy num-
ory with applications, Soft Computing 24(22) (2020), bers and fuzzy risk analysis, Advances in Mathematical
16727–16743. Physics 28 (2015), 1–12.
[65] H. Kamaci, Selectivity analysis of parameters in soft set [68] Z. Zararsiz, A contribution to the algebraic structure of fuzzy
and its effect on decision making, International Jour- numbers, Annals of Fuzzy Mathematics and Informatics
nal of Machine Learning and Cybernetics 11(2) (2020), 12(2) (2016), 205–219.
313–324. [69] Z. Zararsiz and M. Sengonul, On the gravity of center of
[66] E. Aygun and H. Kamaci, Some new algebraic structures of sequence of fuzzy numbers, Annals of fuzzy Mathematics
soft sets, Soft Computing 25(13) (2021), 8609–8626. and Informatics 6(3) (2013), 479–485.

PY
CO
OR
TH
AU

View publication stats

You might also like