Q. Pang et al. / Information Sciences 369 (2016) 128–143 131
where b
ρ(l)
α
and b
ρ(l)
β
are the lth linguistic terms in b
α
and b
β
respectively, # b
α
and # b
β
are the numbers of the linguistic
terms in b
α
and b
β
respectively.
Another operational law can be defined as:
λb
ρ
α
= ∪
b
ρ(l)
α
∈ b
α
{ λb
ρ(l)
α
} , λ ≥ 0 (5)
Theorem 1. [42] . Let b
1
, b
2
and b
3
be any three HFLTSs, # b
1
= # b
2
= # b
3
, λ, λ
1
, λ
2
≥ 0 , where b
ρ(l)
1
, b
ρ(l)
2
and b
ρ(l)
3
are the lth
linguistic terms in b
1
, b
2
and b
3
respectively. Then
(1) b
ρ
1
b
ρ
2
= b
ρ
2
b
ρ
1
;
(2) (b
ρ
1
b
ρ
2
) b
ρ
3
= b
ρ
1
(b
ρ
2
b
ρ
3
) ;
(3) λ(b
ρ
1
b
ρ
2
) = λb
ρ
1
λb
ρ
2
;
(4) ( λ
1
+ λ
2
) b
ρ
1
= λ
1
b
ρ
1
λ
2
b
ρ
1
.
During the computational process, the resultant elements of a HFLTS may not be linguistic terms in S of Eq. (1) but
virtual terms in
¯
S
of Eq. (2) . According to Ref. [26] , a set of finite number of continuous linguistic terms is considered as a
HFLTS as well.
We can see that although HFLTSs allow the DMs to express their opinions by using several linguistic terms, but they
cannot reflect the weights of the DMs’ opinions in group decision making.
3. Probabilistic linguistic term sets
In order to overcome the abovementioned issue of HFLTSs, in this section, we will propose a novel concept called PLTSs,
and investigate the comparison method, the basic operation laws and the aggregation operators.
3.1. The concept of PLTSs
As stated in Section 1 , the DMs may have hesitancy among several possible linguistic terms when expressing their pref-
erences. Moreover, the complete probabilistic distribution on these linguistic terms is usually not so easy to be provided
accurately. In this case, only the known information of probability can be used. As seen in Example 1 , only the lower
bounds of the probabilities of the two linguistic terms are available, or only parts of the probabilistic distribution can be
derived, as in Example 2 . Generally, we extend HFLTSs and other existing techniques by the following definition:
Definition 3. Let S = { s
0
, s
1
, ... , s
τ
} be a LTS, a PLTS can be defined as:
L (p) =
L
(k )
( p
(k )
) | L
(k )
∈ S, p
(k )
≥ 0 , k = 1 , 2 , ..., # L (p) ,
# L (p)
k =1
p
(k )
≤ 1
(6)
where L
(k )
( p
(k )
) is the linguistic term L
(k )
associated with the probability p
(k )
, and # L (p) is the number of all different
linguistic terms in L (p) .
Note that if
# L (p)
k =1
p
(k )
= 1 , then we have the complete information of probabilistic distribution of all possible linguistic
terms; if
# L (p)
k =1
p
(k )
< 1 , then partial ignorance exists because current knowledge is not enough to provide complete as-
sessment information, which is not rare in practical GDM problems. Especially,
# L (p)
k =1
p
(k )
= 0 means completely ignorance.
Obviously, handling the ignorance of L (p ) is a crucial work for the use of PLTSs.
Since the positions of elements in a set can be swapped arbitrarily, then we first propose the ordered PLTSs to ensure
that the operational results among PLTSs can be straightforwardly determined.
Definition 4. Given a PLTS L (p) = { L
(k )
( p
(k )
) | k = 1 , 2 , ..., # L (p) } , and r
(k )
is the subscript of linguistic term L
(k )
. L (p) is called
an ordered PLTS, if the linguistic terms L
(k )
( p
(k )
) ( k = 1 , 2 , ..., # L (p) ) are arranged according to the values of r
(k )
p
(k )
( k =
1 , 2 , ..., # L (p) ) in descending order.
Example 5. Suppose that the linguistic term set S of Example 3 is the set used in Example 2 , then the PLTS can be denoted
by L (p) = { s
4
(0 . 1) , s
5
(0 . 65) , s
6
(0 . 2) } . If the subscripts of linguistic terms in L (p) are multiplied by the corresponding prob-
abilities, then we get 4 × 0 . 1 = 0 . 4 , 5 × 0 . 65 = 3 . 25 , and 6 × 0 . 2 = 1 . 2 . Reordering the linguistic terms in L (p) in descending
order, we have
L (p) = { s
5
(0 . 65) , s
6
(0 . 2) , s
4
(0 . 1) }
3.2. The normalization of PLTSs
There are two distinct tasks in the normalization of PLTSs. The first one is to estimate the ignorance of probabilistic
information, and the second one is to normalize the cardinality of a PLTS for the purpose of computation.