Elsevier

Information Sciences

Volume 280, 1 October 2014, Pages 338-351
Information Sciences

An outranking approach for multi-criteria decision-making with hesitant fuzzy linguistic term sets

https://doi.org/10.1016/j.ins.2014.05.012Get rights and content

Abstract

The main purpose of this paper is to suggest an approach of multi-criteria decision-making that combines a hesitant fuzzy linguistic term set (HFLTS) with an outranking method involving systematic comparisons of the assessment values of alternatives for each criterion. A HFLTS is more suitable than the traditional fuzzy linguistic set for describing decision-makers’ preferences when they prefer to use several linguistic terms in expressions. After reviewing existing studies of HFLTSs, a novel outranking relational system of HFLTSs is developed and dominance relations for HFLTSs are introduced using outranking degrees of HFLTSs. In order to verify the efficacy of the developed method, an illustrative example is given and comparisons are made between the proposed method and other existing ones.

Introduction

In practice, multi-criteria decision-making (MCDM) methods are widely used to rank alternatives or select the best one with respect to several concerned criteria. In classical MCDM methods, the assessments of alternatives are precisely known [14], [45]. Due to the fuzziness and uncertainty of decision-making problems and the inherent vagueness of human preferences, however, the best expression of decision-makers comes in natural language. That is why using linguistic assessments is much more realistic than using numerical values. Therefore, linguistic variables [68] whose values are words or sentences from natural or artificial languages are widely used in assessing alternatives with respect to criteria.

In recent years, MCDM problems with criteria values derived from linguistic terms have attracted much attention of the researchers [9], [17], [18], [21], [27], [35], [46], [52], [53], [54], [60] in which the semantics of linguistic terms are usually given priority on a qualitative scale. This type of priority is more expressive for decision-makers than a simple order because of the presence of absolute landmarks [13]. One of the four main schemes in aggregating such linguistic information uses operations that are performed with associated membership functions [8], [25] and a variety of aggregation operators for different fuzzy sets [4], [71]. This scheme transforms linguistic information into fuzzy sets such as triangular fuzzy numbers, trapezoidal fuzzy numbers or type-2 fuzzy sets [30], [31] and several operators such as type-1 OWA and type-2 OWA operators have been recently introduced using this scheme [70], [72]. The second scheme involves the direct use of linguistic labels and requires a demand for computing with words (CWW) and several aggregation operators have been developed to meet the demand [5], [10], [20], [65]. The third scheme deals with linguistic information based on a model with 2-tuple fuzzy linguistic representation [21], with some researchers proposing extended 2-tuple fuzzy linguistic models [11], [12], [19], [47] and developing a variety of 2-tuple linguistic aggregation operators [50], [56], [57], [58], [61], [67]. The fourth scheme explores the application of the related theories in transforming linguistic information. The cloud model, a representative of these theories, can correctly depict the uncertainty of a qualitative concept. This model has been successfully utilized for handling uncertain linguistic variables [48] and Atanassov’s interval-valued intuitionistic linguistic numbers [51].

The aforementioned approaches have a common characteristic in which each linguistic variable is expressed by a single linguistic term. However, because of uncertainty a single term is not suitable or sufficient for decision-makers to express their preferences or opinions. Therefore, some investigations have focused on decision-making in uncertain linguistic environments, and use uncertain linguistic variables that are actually interval variables to handle uncertainty while others have used uncertain linguistic aggregation operators and methods for multi-criteria group decision-making (MCGDM) problems [62]. Additionally, more aggregation operators for uncertain linguistic variables and their application in decision-making can be found in other investigations [34], [36], [49], [63], [64], [66]. An uncertain linguistic variable is continuous in a linguistic interval, which means that the probabilities of all values in this interval are equal, or follow a specified distribution. However, decision-makers may deem that linguistic intervals cannot reflect their preferences in fact so they prefer to use multiple terms for better expression.

Hesitant fuzzy linguistic term sets (HFLTSs) [37], [38], [40] were proposed and used to deal with the situations where decision-makers think of several possible linguistic values or richer expressions than a single linguistic term for an indicator, alternative, variable, etc. Compared to other fuzzy linguistic approaches, HFLTSs are more convenient and flexible to reflect decision-makers’ preferences. Recently, the studies on MCDM problems with HFLTSs are growing. For example, Lee and Chen [24] proposed a new fuzzy decision-making method based on likelihood-based comparison relations of HFLTSs, while Rodríguez et al. [39] proposed a new linguistic group decision model that facilitates the elicitation of flexible and rich linguistic expressions. Zhu and Xu [73] proposed a hesitant fuzzy linguistic preference relation and its consistency measures based on HFLTSs while Beg and Rashid [3] proposed a fuzzy TOPSIS method to aggregate the opinion of experts or decision-makers in MCGDM problems, where the opinions of experts are represented by HFLTSs. Wei et al. [55] defined the operations and two aggregation operators for HFLTSs, and then proposed the decision-making methods to deal with MCDM problems. Liao et al. [26] gave the distance and similarity measures for HFLTSs and applied them in MCDM problems while the fuzzy envelope of an HFLTS [28] was constructed using a fuzzy membership function and then combined with the fuzzy TOPSIS model to solve MCDM problems. Furthermore, Zhang and Wu [69] extended HFLTSs and introduced hesitant fuzzy linguistic sets (HFLSs) that can contain inconsecutive linguistic terms. Also, hesitant fuzzy linguistic aggregation operators were developed to aggregate the assessment values in the form of HFLTSs. Finally, Meng et al. [32] defined linguistic hesitant fuzzy sets (LHFSs) that considered the possible membership degrees of each linguistic term. The same group developed aggregation operators that could be applied in MCDM problems under a linguistic hesitant fuzzy environment.

Although function models have been adopted by most of the existing methods based on HFLTSs, using aggregation operators may bring some drawbacks in operations. (1) If the aggregation operators of HFLTSs were used, the defined discrete linguistic values would be necessarily extended to the continuous ones as in a simple operation using the subscripts of linguistic terms: 0.5  {s1}  0.5  {s2} = {s1.5}. In such cases, there is an awareness that s1.5 does not have any syntax or semantics assigned, because such a virtual linguistic term makes sense only in comparison and in operation. Nevertheless, the existence of s1.5 may affect the effectiveness of HFLTSs to some extent because HFLTSs would be the same as uncertain linguistic variables in this situation. By contrast, the use of the relation models can avoid the appearance of virtual linguistic terms which are unavoidable in the case of using the aggregation operators. (2) Furthermore, the union of two HFLTSs may result in a set containing inconsecutive linguistic terms, e.g., {s1}  {s5} = {s1,s5}, which contradicts the definition of HFLTSs and (3) the results using aggregation operators are possibly not consistent with common sense. For example, Wei et al. [55] discussed that in Rodríguez’s comparison results [38] it seems to be unreasonable to declare one HFLTS absolutely superior to another if two HFLTSs have common elements. Therefore, an outranking method ought to be recommended to handle MCDM problems with HFLTSs.

All outranking methods permit comparability and intransitivity of preferences [42] and enable the utilization of incomplete information such as judgments on ordinal measurement scales and partial prioritization. Furthermore, they are able to take ordinal scales into account without converting them into abstract ones with an arbitrary imposed range [43] while maintaining the original verbal meanings [16]. Meanwhile, the thresholds can be considered when modeling imperfect knowledge [29]. Outranking methods, including ELECTRE I, ELECTRE II, ELECTRE III, etc., are important methods in MCDM problems, and have been widely used [74]. In these methods, all couples of alternatives are respectively compared under each criterion before determining the preferred one. Among all outranking methods, ELECTRE I [41] and its derivatives [15] are most prominent and have been successfully used in many fields [1], [2], [6], [7], [22], [23], [33], [44], [67].

As mentioned previously, the outranking method is one of the most appropriate methods to solve MCDM problems because of its simple logic. Nevertheless, the literature shows that none of current studies use an outranking method to solve MCDM problems with HFLTSs. Furthermore, many previous studies on outranking methods have focused on certain data while our research is based on uncertain data as an extension of outranking methods.

This paper focuses on how to solve an MCDM problem using HFLTSs, and the proposed approach is an integration of both HFLTSs and ELECTRE I. The assessments of alternatives with respect to criteria are expressed in the form of HFLTSs, and the outranking relations between alternatives are determined by systematically comparing the assessment values for each criterion. The rest of the paper is organized as follows. In Section 2, the concepts and operations of HFLTSs are briefly reviewed together with a distance for HFLTSs. In Section 3, the dominance relations between HFLTSs and some useful properties are defined. Subsequently, the concordance and discordance indices are put forward and an outranking approach for MCDM problems using HFLTSs is developed in Section 4. In Section 5, the proposed approach is demonstrated by using an illustrative example and the results are compared to those using other existing methods. Finally, conclusions are drawn in Section 6.

Section snippets

Preliminaries

In this section, some concepts and operations of HFLTSs are briefly reviewed, and then a distance for HFLTSs is introduced.

Let S = {sii = 0,  , g,g  N} be a linguistic term set, where si represents a possible value for a linguistic variable, and the following characteristics should be satisfied:

  • (1)

    The set is ordered: if α > β, then sα > sβ.

  • (2)

    There exists a negation operator: neg(sα) = sgα.

  • (3)

    If sα > sβ, then max{sα, sβ} = sα and min{sα,sβ} = sβ.

The outranking method based on HFLTSs

In this section, the dominance relations between HFLTSs are put forward and some desirable properties are explored to support these definitions. Then, in order to quickly determine the dominance relations between HFLTSs, some propositions are proposed.

Definition 6

Let S = {s0, s1,  , sg} be a linguistic term set and si and sj be two arbitrary linguistic terms on S. A binary relation P can be then defined as follows:P(si,sj)=1ifsi>sj0ifsisj

Definition 7

Let S = {s0, s1,  , sg} be a linguistic term set, HS1 and HS2 be two arbitrary

An MCDM outranking approach based on HFLTSs

This section describes a new MCDM approach for decision-making by integrating HFLTSs and the outranking method with evaluation information.

In a multi-criteria linguistic ranking or selection problem, there are n alternatives, denoted by A = {a1, a2,  , an}. Each alternative is assessed by means of m criteria, denoted by C = {c1, c2,  , cm}, and the evaluations are undertaken using linguistic expressions. The weight of the criterion cj is wj, where wj  0 (j = 1, 2,  , m), and j=1mwj=1.

In this decision-making

Illustrative example

As is known, one of the important capabilities in information warfare is the offensive warfare capability – the ability of interfering and damaging the hostile information system. The offensive warfare capability may be decomposed into four evaluation criteria – the capability of information suppression C1, the capability of hard destruction C2, the capability of network resistance C3 and the capability of psychological resistance C4. Assuming the weight vector of the four criteria is W = (0.3, 

Conclusions

A HFLTS, defined as an ordered finite subset of consecutive linguistic terms, is more suitable than traditional fuzzy linguistic approaches when experts hesitate among several values to assess a linguistic variable. However, while outranking methods are well-established MCDM methods and their application has shown considerable success in practice, little attention has been paid to the combination of outranking methods with HFLTSs. The main purpose of this paper is to call attention to the

Acknowledgements

The authors thank the editors and anonymous reviewers for their helpful comments and suggestions. This work was supported by the National Natural Science Foundation of China (Nos. 71221061 and 71271218) and the Hunan Provincial Natural Science Foundation of China (No. 14JJ2009).

References (74)

  • J. Hokkanen et al.

    Choosing a solid waste management system using multicriteria decision analysis

    Eur. J. Oper. Res.

    (1997)
  • D.F. Li et al.

    Linear programming method for multiattribute group decision making using IF sets

    Inf. Sci.

    (2010)
  • H.C. Liao et al.

    Distance and similarity measures for hesitant fuzzy linguistic term sets and their application in multi-criteria decision making

    Inf. Sci.

    (2014)
  • L.Z. Lin et al.

    Fuzzy linguistic decision to provide alternatives to market mechanism strategies

    Expert Syst. Appl

    (2010)
  • H.B. Liu et al.

    A fuzzy envelope for hesitant fuzzy linguistic term set and its application to multicriteria decision making

    Inf. Sci.

    (2014)
  • F.Y. Meng et al.

    Multi-attribute decision analysis under a linguistic hesitant fuzzy environment

    Inf. Sci.

    (2014)
  • A. Papadopoulos et al.

    Application of the multi-criteria analysis method Electre III for the optimisation of decentralised energy systems

    Omega

    (2008)
  • R.O. Parreiras et al.

    A flexible consensus scheme for multicriteria group decision making under linguistic assessments

    Inf. Sci.

    (2010)
  • B. Peng et al.

    Uncertain pure linguistic hybrid harmonic averaging operator and generalized interval aggregation operator based approach to group decision making

    Knowl. Based Syst.

    (2012)
  • R.M. Rodríguez et al.

    A group decision making model dealing with comparative linguistic expressions based on hesitant fuzzy linguistic term sets

    Inf. Sci.

    (2013)
  • A. Shanian et al.

    A new application of ELECTRE III and revised Simos’ procedure for group material selection under weighting uncertainty

    Knowl. Based Syst.

    (2008)
  • T.J. Stewart

    A critical survey on the status of multiple criteria decision making theory and practice

    Omega

    (1992)
  • T.C. Wang et al.

    Applying fuzzy linguistic preference relations to the improvement of consistency of fuzzy AHP

    Inf. Sci.

    (2008)
  • J.Q. Wang et al.

    Method of multi-criteria group decision-making based on cloud aggregation operators with linguistic information

    Inf. Sci.

    (2014)
  • W.P. Wang

    Evaluating new product development performance by fuzzy linguistic computing

    Expert Syst. Appl.

    (2009)
  • W.P. Wang

    A fuzzy linguistic computing approach to supplier evaluation

    Appl. Math. Modelling

    (2010)
  • G.W. Wei

    A method for multiple attribute group decision making based on the ET-WG and ET-OWG operators with 2-tuple linguistic information

    Expert Syst. Appl.

    (2010)
  • G.W. Wei

    Some generalized aggregating operators with linguistic information and their application to multiple attribute group decision making

    Comput. Ind. Eng.

    (2011)
  • G.W. Wei et al.

    Some dependent aggregation operators with 2-tuple linguistic information and their application to multiple attribute group decision making

    Expert Syst. Appl.

    (2012)
  • M.C. Wu et al.

    The ELECTRE multi-criteria analysis approach based on Atanassov’s intuitionc fuzzy sets

    Expert Syst. Appl.

    (2011)
  • Y.J. Xu et al.

    A method for multiple attribute decision making with incomplete weight information under uncertain linguistic environment

    Knowl. Based Syst.

    (2008)
  • Y.J. Xu et al.

    Approach based on 2-tuple linguistic power aggregation operators for multiple attribute group decision making under linguistic environment

    Appl. Soft Comput.

    (2011)
  • Z.S. Xu

    Uncertain linguistic aggregation operators based approach to multiple attribute group decision making under uncertain linguistic environment

    Inf. Sci.

    (2004)
  • Z.S. Xu

    An approach based on the uncertain LOWG and the induced uncertain OWG operators to group decision making with uncertain mutiplicative linguistic preference relations

    Decis. Support Syst.

    (2006)
  • Z.S. Xu

    Induced uncertain linguistic OWA operators applied to group decision making

    Inf. Fusion

    (2006)
  • W. Yang et al.

    New aggregation operators based on the Choquet intergral and 2-tuple linguistic information

    Expert Syst. Appl.

    (2012)
  • L. Zadeh

    The concept of a linguistic variable and its application to approximate reasoning—I

    Inf. Sci.

    (1975)
  • Cited by (157)

    View all citing articles on Scopus
    View full text