expressions directly. Hence, the introduction of HFLTS is a worth-
while contribution to the theory of CWW.
As the HFLTS increases the flexibility and capability of the
elicitation of linguistic information by means of linguistic expres-
sions, it has attracted more and more scholars’ attention and
many fruitful achievements have been proposed. After introduc-
ing the concept of HFLTS and its basic operations and properties,
Rodríguez et al. (2012) developed a multi-criteria linguistic
decision making model with linguistic expressions. Later on,
they (Rodríguez, Martínez, & Herrera, 2013) further proposed a
linguistic group decision making model which deals with
comparative linguistic expressions in the process of group deci-
sion making. To make it much easier to understand, Liao, Xu,
Zeng, and Merigó (2015c) gave the mathematical definition of
HFLTS and introduced the hesitant fuzzy linguistic element
(HFLE) to represent the hesitant fuzzy linguistic value of a
linguistic variable. Liao, Xu, and Zeng (2014b), Liao et al.
(2015c) also introduced different kinds of distance measures,
similarity measures and correlation coefficients for HFLTSs. Wei,
Zhao, and Tang (2014) investigated the comparison and aggrega-
tion methodologies for HFLTSs. Based on the fuzzy envelope, a
new representation of the HFLTS was introduced by Liu and
Rodríguez (2014). In order to apply the HFLTS into decision
making process, motivated by the hesitant fuzzy preference
relation (Liao, Xu, & Xia, 2014a), Zhu and Xu (2014) defined the
hesitant fuzzy linguistic preference relation (HFLPR) and investi-
gated its consistency. Liu, Cai, and Jiang (2014) further investi-
gated the additive consistency of a HFLPR. Some decision
making approaches were also extended into HFLTS circumstances.
For example, for the multi-criteria decision making (MCDM)
problem in which the opinions of an expert are represented by
HFLTSs, Beg and Rashid (2013) proposed a TOPSIS-based method
to solve it. As to the hesitant fuzzy linguistic MCDM problem
where some criteria conflict with each other, Liao, Xu, and Zeng
(2015a) established a HFL-VIKOR method and implemented it
into practical decision making processes.
All the above-mentioned results show that the HFLTS is a good
research topic in the field of decision making. However, as the
HFLTS is just introduced in 2012, the foundation of this theory is
not strong enough and thus much work needs to be done to fill this
gap. Roughly speaking, the achievements on HFLTSs in the litera-
ture can be classified into the following parts:
Basic definitions and operations over HFLTSs (see Liao et al.
(2015c), Liu et al. (2014), Liu & Rodríguez (2014), Rodríguez
et al. (2012), Zhu & Xu (2014));
Information fusion methods with HFLTSs, such as different
forms of aggregation operators (see Rodríguez et al. (2012),
Wei et al. (2014));
Measures of HFLTSs, including the correlation measures (Liao
et al., 2015c), the distance measures, and the similarity mea-
sures ( Liao et al., 2014b);
Distinct decision making methods, such as the group decision
making method (Rodríguez et al., 2013), the HFL-TOPSIS
method (Beg & Rashid, 2013; Liu & Rodríguez, 2014), and the
HFL-VIKOR method (Liao et al., 2015a).
Since both the HFL-TOPSIS method and the HFL-VIKOR
method are based on the distance measures of HFLTSs, in this
paper, we focus our attention on the distance and similarity
measures of HFLTSs. The basic principle of the TOPSIS method
is to find an alternative which has the shortest distance from
the positive-ideal solution and the furthest distance from the
negative-ideal solution (Beg & Rashid, 2013; Hwang & Yoon,
1981); while the main idea of VIKOR method is to determine a
compromise solution, which provides a maximum ‘‘group utility’’
for the ’’majority’’ and a minimum ‘‘individual regret’’ for the
‘‘opponent’’, for a MCDM problem with non-commensurable and
conflicting criteria by mutual concessions (Opricovic & Tzeng,
2004), but such ‘‘group utility’’ and ‘‘individual regret’’ are
measured by the distances measures of HFLTSs. Hence, the
distance measures of the HFLTSs play a very important role in
these two methods. As we can see from Liao et al. (2014b) that
all the distance measures proposed by them were based on the
different forms of algebra distance measures, such as the
Hamming distance measure, the Euclidean distance measure
and the Hausdorff distance measure. In this paper, we try to
propose some novel distance measures which are not based on
the algebraic distance measures but from the geometric point
of view. A sort of cosine distance and similarity measures are
introduced for HFLTSs, based on which, the cosine-distance-
based HFL-TOPSIS and the cosine-distance-based HFL-VIKOR
method are further established. A numerical example concerning
the selection of ERP system is given to illustrate the validation
and efficiency of the proposed method.
The remainder of this paper is organized as follows: Section 2
reviews the concepts of HFLTS and the distance and similarity
measures. Section 3 proposes different forms of cosine distance
and similarity measures for HFLTSs. The cosine-distance-based
HFL-TOPSIS method and the cosine-distance-based HFL-VIKOR
method are developed in Section 4. A numerical example is given
in Section 5 to show the applicability and validation of the method-
ologies. The paper ends with some concluding remarks in
Section 6.
2. Hesitant fuzzy linguistic term set and the distance and
similarity measures
2.1. Hesitant fuzzy linguistic term set
In traditional fuzzy linguistic approach for qualitative decision
making, experts are supposed to use single linguistic term to
represent the value of a linguistic variable. However, this is not
appropriate to tackle more complicated decision making problems
as in many cases, the experts can not give their assessments in
single terms but linguistic expressions. For example, when eval-
uating the performance of an operation system, one engineer
may say ‘‘its performance is between medium and high’’, while
the other may deem ‘‘it is at least a little high’’. As traditional fuzzy
linguistic approach can only use single term, such as ‘‘medium’’,
‘‘high’’ or ‘‘a little high’’, to express the cognition of a person, in
order to represent comprehensive linguistic expressions,
Rodríguez et al. (2012) introduced the concept of hesitant fuzzy
linguistic term set, which can be used to elicit several linguistic
terms for a linguistic variable.
Definition 1 Rodríguez et al., 2012. Let S ¼fs
0
; ...; s
s
g be a
linguistic term set. A hesitant fuzzy linguistic term set (HFLTS),
H
S
, is an ordered finite subset of the consecutive linguistic terms
of S.
Example 1. Here we just consider a simple example that an expert
evaluates the operational complexity of three automatic systems,
represented as x
1
; x
2
; andx
3
. Since this criterion is qualitative, it is
impossible to give crisp values but only linguistic terms. The opera-
tional complexity of these automatic systems can be taken as a lin-
guistic variable. The linguistic term set of the operational
complexity can be set up as:
S ¼ s
3
¼
v
ery complex; s
2
¼ complex; s
1
¼ a little complex;
f
s
0
¼ medium; s
1
¼ a little easy; s
2
¼ easy; s
3
¼
v
ery easy
g
H. Liao, Z. Xu / Expert Systems with Applications 42 (2015) 5328–5336
5329