Definition 2.7 [16]. An intuitionistic fuzzy set A in X defined by
Atanassov can be written as:
A ¼ x;
l
A
ðxÞ;
v
A
ðxÞ
x 2 X
j
ð9Þ
where
l
A
ðxÞ : X !½0; 1 and
v
A
ðxÞ : X !½0; 1 are membership
degree and non-membership degree, respectively, with the
condition:
0 6
l
A
ðxÞþ
v
A
ðxÞ 6 1 ð10Þ
p
A
ðxÞ determined by:
p
A
ðxÞ¼1
l
A
ðxÞ
v
A
ðxÞð11Þ
is called the hesitancy degree of the element x 2 X to the set A, and
p
A
ðxÞ2½0; 1; 8x 2 X.
p
A
ðxÞ is also called the intuitionistic index of x to A. Greater
p
A
ðxÞ indicates more vagueness on x. Obviously, when
p
A
ðxÞ¼0;
8
x 2 X, the IF set degenerates into an ordinary fuzzy
set. In the sequel, the couple
l
A
ðxÞ;
v
A
ðxÞ
is called an IF set or IF
event for clarity. IFSðXÞ denotes the family of all IF sets in X.
Definition 2.8 [36]. For A 2 IFSðXÞ and B 2 IFSðXÞ, some operations
on them are defined as:
(1) A B iff
8
x 2 X
l
A
ðxÞ 6
l
B
ðxÞ;
v
A
ðxÞ P
v
B
ðxÞ,
(2) A ¼ B iff
8
x 2 X
l
A
ðxÞ¼
l
B
ðxÞ;
v
A
ðxÞ¼
v
B
ðxÞ,
(3) A
C
¼ x;
v
A
ðxÞ;
l
A
ðxÞ
x 2 X
j
, where A
C
is the complemen-
tary set of A.
(4) A \ B ¼ x; min
l
A
ðxÞ;
l
B
ðxÞ
; max
c
A
ðxÞ;
c
B
ðxÞðÞ
x 2 X
j
,
(5) A [ B ¼ x; max
l
A
ðxÞ;
l
B
ðxÞ
; min
c
A
ðxÞ;
c
B
ðxÞ
ðÞ
x 2 X
j
,
(6) A þB ¼ x;
l
A
ðxÞþ
l
B
ðxÞ
l
A
ðxÞ
l
B
ðxÞ;
c
A
ðxÞ
c
B
ðxÞ
x 2 X
j
,
(7) A B ¼ x;
l
A
ðxÞ
l
B
ðxÞ;
c
A
ðxÞþ
c
A
ðxÞ
c
A
ðxÞ
c
B
ðxÞ
x 2 X
j
.
It is worth noting that besides Definition 2.7, there are other
possible representations of IF sets proposed in the literature.
Hong and Choi [37] proposed to use an interval representation
l
A
ðxÞ; 1
v
A
ðxÞ
of intuitionistic fuzzy set A in X instead of pair
l
A
ðxÞ;
v
A
ðxÞ
. This approach is equivalent to the interval valued
fuzzy sets interpretation of IF set, where the interval
l
A
ðxÞ; 1
v
A
ðxÞ
represents the membership degree of x 2 X to
the set A. Obviously,
l
A
ðxÞ; 1
v
A
ðxÞ
is a valid interval, since
l
A
ðxÞ 6 1
v
A
ðxÞ always holds for
l
A
ðxÞþ
v
A
ðxÞ 6 1.
2.3. Intuitionistic fuzzy evidence theory
It has been well known that Dempster–Shafer evidence theory
can express and deal with uncertainty in crisp sets. However, D-S
theory itself cannot represent and manage vague information such
as ‘‘the price is high’’ or ‘‘his age is about 40’’. To overcome this
problem, fuzzy evidence theory was proposed to process imprecise
and vague information [18–24].
Definition 2.9. Let X ¼fx
1
; x
2
; ...; x
n
g be a universe of discourse,
and FSðXÞ be the family of all fuzzy sets in X. A fuzzy belief function
m is defined as:
A
F
i
; mA
F
i
;
l
A
F
i
ðx
j
Þ
DEno
; A
F
i
2 FSðXÞ; x
j
2 X ð12Þ
where A
F
i
is a normal fuzzy set,
l
A
F
i
ðx
j
Þ is the membership function
of the fuzzy set A
F
i
, and mA
F
i
is the BPA of A
F
i
A fuzzy set A
F
2 FSðXÞ
with mðA
F
Þ > 0 is referred to as a focal element of m.
Definition 2.10 [18]. Let the probability of each element in X be
Pðx
j
Þðj ¼ 1; 2; ...; nÞ. The BPA of a fuzzy event
A
F
i
¼ x
j
;
l
A
F
i
ðx
j
Þ
DE
x
j
2 X
no
is defined as:
mA
F
i
¼
X
n
j¼1
P ðx
j
Þ
l
A
F
i
ðx
j
Þð13Þ
Definition 2.11. Let X ¼fx
1
; x
2
; ...; x
n
g be a universe of discourse,
and IFSðXÞ be the family of all fuzzy sets in X. An intuitionistic fuzzy
belief function m is defined as:
A
IF
i
; mA
IF
i
;
l
A
IF
i
ðx
j
Þ;
v
A
IF
i
ðx
j
Þ
DEno
; A
IF
i
2 IFSðXÞ; x
j
2 X ð14Þ
where A
IF
i
is a is an intuitionistic fuzzy set,
l
A
IF
i
ðx
j
Þ is the member-
ship function of the fuzzy set A
IF
i
;
v
A
IF
i
ðx
j
Þ is the membership function
of the intuitionistic fuzzy set A
IF
i
, and mA
IF
i
is the BPA of A
IF
i
.
Definition 2.12 [25]. Let the probability of each element in X be
Pðx
j
Þ (j ¼ 1; 2; ...; n). The BPA of an IF event
A
IF
i
¼ x
j
;
l
A
IF
i
ðx
j
Þ;
v
A
IF
i
ðx
j
Þ
DE
x
j
2 X
no
is an interval value defined as:
mA
IF
i
¼ m
min
A
IF
i
; m
max
A
IF
i
hi
ð15Þ
where m
min
A
IF
i
and m
max
A
IF
i
are the minimal and maximal BPA
of A
IF
i
, respectively, defined as:
m
min
A
IF
i
¼
X
n
j¼1
Pðx
j
Þ
l
A
IF
i
ðx
j
Þð16Þ
m
max
A
IF
i
¼
X
n
j¼1
Pðx
j
Þð1
v
A
IF
i
ðx
j
ÞÞ ¼ m
min
A
IF
i
þ
X
n
j¼1
P ðx
j
Þ
p
A
IF
i
ðx
j
Þ
ð17Þ
It is easy to verify 0 6 m
min
A
IF
i
6 m
max
A
IF
i
6 1. We shall take
the example analyzed in [39] to show the computation of BPA for
IF events.
Example 2.1. Suppose there is a discussion on the amount of
money which should be assigned for advertisement of a new prod-
uct. Let us assume that the IF event A corresponding to the state-
ment ‘‘about fifty thousands’’ is given by
A ¼ 30; 0:2; 0:6
hi
; 40; 0:5; 0:4
hi
; 50; 1; 0
hi
; 60; 0:5; 0:4
hi
; 70; 0:2; 0:6
hifg
The probability distribution of the amount of money is
P ð30Þ¼Pð40Þ¼Pð50Þ¼Pð60Þ¼Pð70Þ¼0:2
From (16) and (17) we get immediately
m
min
¼ 0:2ð 0: 2 þ 0: 5 þ 1 þ 0:5 þ 0:2Þ¼0:48
m
max
¼ 0:2ð0:4 þ 0:6 þ 1 þ 0:6 þ 0:4Þ¼0:6
It means that the basic probability assigned to the event of ‘‘as-
signing about fifty thousands for advertisement’’ lies in the interval
[0.48, 0.6].
Definition 2.13. Let F denote the set of all focal elements (includ-
ing crisp sets, fuzzy sets or IF sets), and Fjjdenote the cardinality of
F. A belief function m is a normalized belief function if it satisfies
X
A
i
2F
mðA
i
Þ¼1 ð18Þ
290 Y. Song et al. / Knowledge-Based Systems 86 (2015) 288–298