Abstract
Single-valued trapezoidal neutrosophic numbers (SVTNNs) have a strong capacity to depict uncertain, inconsistent, and incomplete information about decision-making problems. Preference relations represent a practical tool for presenting decision makers’ preference information regarding various alternatives. The purpose of this paper is to propose single-valued trapezoidal neutrosophic preference relations (SVTNPRs) as a strategy for tackling multi-criteria decision-making problems. First, this paper briefly reviews basic concepts about neutrosophic sets and SVTNNs and defines a new comparison method and new operations for SVTNNs. Next, two aggregation operators, the single-valued trapezoidal neutrosophic weighted arithmetic average operator and the single-valued trapezoidal neutrosophic weighted geometric average operator, are proposed for applications in information fusion. Then, this paper discusses the definitions of completely consistent SVTNPRs and acceptably consistent SVTNPRs. Finally, we outline a decision-making method based on SVTNPRs to address green supplier selection problems, and we conduct a comparison study and discussion to illustrate the rationality and effectiveness of the decision-making method.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
1 Introduction
Recently, growing concerns about environmental issues have attracted worldwide attention to innovative business practices that alleviate or prevent negative environmental effects [1]. One potentially effective way of managing a company’s environmental policy is by linking it closely with its purchasing function activities, i.e., through supplier selection [2]. Taking the suppliers’ environmental performance into consideration, organizations and governments have attached great value to green supply chain management (GSCM). The processes of green supplier evaluation and selection are critical issues in GSCM [3], because incorporating environmental criteria into these processes can contribute to achieving GSCM goals [4]. Thus, it is critical and necessary to study green supplier evaluation and selection problems.
Green supplier selection problems involve strategic and complex decision making that demands consideration of different criteria, such as green products, green knowledge transfer, and environmental management systems [3]. Selecting the appropriate green supplier can have a direct impact on the reduction in enterprise costs, increase in enterprise flexibility, and the promotion of core competitiveness. Such complex problems can be solved using multi-criteria decision-making (MCDM) techniques, which can not only facilitate reaching clear decisions but also deal with various, often conflicting criteria [5]. MCDM techniques are very useful tools for addressing many real-life green supplier selection problems.
To the best of our knowledge, existing research into green supplier evaluation and selection can roughly be classified into the following MCDM approaches: the analytic hierarchy process (AHP) [6, 7], the analytic network process (ANP) [8–10], mathematical programming [11, 12], and some other approaches [13–16]. Because decision makers (DMs) often have limited time and knowledge, the vagueness of their opinions must also be taken into consideration; for this reason, fuzzy logic and fuzzy sets (FSs), initially proposed by Zadeh [17], have been integrated in the methods listed above. For example, Chan et al. [18] incorporated fuzzy logic into the fuzzy-AHP model to measure the environmental and organizational performance of different designs for eco-friendly products. Kumar et al. [19] used an integrated fuzzy-AHP and fuzzy multi-objective linear programming approach for order allocation among green suppliers. AHP and fuzzy-AHP are recognized as good tools for addressing MCDM problems, as they can provide techniques for flexibly deciding among various options [20]. However, real-world complexity has generated the need to extend other fuzzy concepts to help organizations make more thoughtful and precise decisions. An extension of FSs, intuitionistic fuzzy sets (IFSs) [21] have demonstrated a strong ability to represent vagueness and uncertainty, such that they can describe MCDM problems in a more accurate way. Büyüközkan and Güleryüz [20] integrated the intuitionistic fuzzy analytic hierarchy process (IF-AHP) and intuitionistic fuzzy technique for order preference by similarity to ideal solution (IF-TOPSIS) methods to effectively evaluate product development partners. Furthermore, interval type-2 fuzzy sets (IT2FSs) [22], which are another extension of FSs, are remarkably flexible in modeling the uncertainty of MCDM problems. Within environments characterized by IT2FSs, researchers made some significant discoveries; for example, Yu et al. [23] proposed a new multi-attributive border approximation area comparison method to solve hotel selection problems in a tourism websites. Ghorabaee et al. [24] extended the Vlsekriterijumska Optimizacija I Kompromisno Resenje (VIKOR) method for handling fuzzy multi-criteria group decision-making (MCGDM) problems using IT2FSs. In another study, Ghorabaee et al. [25] presented a new method for ranking interval type-2 fuzzy numbers (IT2FNs) and extended the complex proportional assessment method for supplier selection problems. Later, in order to consider environmental criteria, Ghorabaee et al. [4] proposed an integrated approach based on the weighted aggregated sum product assessment method to deal with green supplier selection problem using IT2FSs.
Although prior studies have contributed to advancing the study of green supplier evaluation and selection problems, these problems can feature incomplete and inconsistent information that remains beyond the scope of FSs, IFSs or other extensions [26–28]. In order to cope with indeterminate and inconsistent information to the greatest extent possible, Smarandache [29, 30] proposed neutrosophic sets (NSs) from a philosophical point of view [31]. Recently, Smarandache [32] extended NSs to propose refined NSs, introducing for the first time the degree of dependence or independence of (sub)components of NSs. However, NSs cannot be applied in real scientific and engineering areas since their description is not specified. To meet this critical challenge, multiple researchers have studied extensions of NSs, including simplified neutrosophic sets [33, 34], multi-valued neutrosophic sets (MVNSs) [35–37], and independent inputs simplified neutrosophic linguistic sets (SNLSs) [38]. Up to this point, these extensions have been applied to MCDM problems to remarkable effect [1]. However, the domains of these extensions are still discrete sets; for example, in independent inputs SNLSs, the three membership degrees are relative to a discrete fuzzy concept “excellent” or “good.” Naturally, this may lead to the loss of information, such that it is worthwhile to extend the discrete set to a continuous one.
To tackle issues analyzed above, Ye [39] proposed single-valued trapezoidal neutrosophic sets (SVTNSs) as another extension of NSs, while Deli and Şubaş [40] defined single-valued triangular neutrosophic numbers (SVTrN-numbers), which can be regarded as special cases of single-valued trapezoidal neutrosophic numbers (SVTNNs). As these studies demonstrate, information from DMs can be expressed in different dimensions and continuous sets. Other researchers have also performed significant work in this field. For example, both Deli and Şubaş [41] and Biswas et al. [42] proposed a new ranking method by defining the concept of cut sets for SVTNNs, which they applied to tackle MCDM problems. Thamaraiselvi and Santhi [43] introduced the mathematical representation of a transportation problem in the SVTNN environment. Based on the extent studies, Liang et al. [44] improved the existing operations and operations of SVTNNs and proposed a new MCGDM approach based on interdependent inputs of SVTNN information. In other fields, however, very little researches exist based on SVTNNs. Considering the positive characteristics of SVTNNs in representing incomplete and inconsistent information, this paper explores further applications of SVTNNs in green supplier selection problems.
In real-life MCDM problems, it seems more flexible for DMs to offer comparisons among alternatives rather than providing evaluation values for all alternatives with respect to each criterion. Moreover, it has been pointed out that building a preference relation by pairwise comparisons between alternatives is more accurate than non-pairwise comparisons [45]. Preference relations fall into two main categories: multiplicative preference relations [46, 47] and fuzzy preference relations (FPRs) [48–50]. Saaty [51] proposed the traditional AHP method based on multiplicative preference relations, using the 1 through −9 linguistic scale to represent judgments made by pairwise comparisons between alternatives, in which all judgments are crisp values. However, deterministic values cannot reflect the vagueness of real-life decision-making information. As a result, Orlovsky [50] introduced FPRs, in which each element has a membership degree assigned from 0 to 1. FSs cannot take the non-membership into consideration, while IFSs can cover this deficiency; therefore, Xu [52] constructed intuitionistic fuzzy preference relations (IFPRs). Using this type of representation technique, DMs can express their imprecise cognitions from positive, negative, and hesitant points of view when comparing alternatives \(A_{i}\) and \(A_{j}\) [53]. Although IFPRs can express DMs’ preferences in a more comprehensive and flexible way than FPRs [53], some weaknesses still exist in IFPRs that go beyond the capacity of intuitionistic fuzzy numbers (IFNs) to handle inconsistent information, which might be common in complex decision-making situations.
To overcome the defects described above, this paper proposes single-valued trapezoidal neutrosophic preference relations (SVTNPRs). In real-life decision-making situations, SVTNPRs are more widespread and can cover more DM preference information than FPRs and IFPRs. For example, suppose that teams are assembled from several constituencies to serve as DMs (reviewers) in order to solve a green supplier selection problem; then, data are collected over two sessions. First, experts are asked to offer their preference degree between each pair of alternatives \(A_{i}\) and \(A_{j}\) \(\left( {i,j = 1,2, \ldots ,m} \right)\) with respect to each criterion. Due to the limited knowledge and indeterminacy inherent in the DMs’ cognitions, they are more likely to use linguistic information rather numeric values to denote their preference values; for example, when asked about the performance of one car, the reviewers might tend to describe the performance as “good” or “poor.” However, based on the preceding discussion about independent inputs SLNSs, it is worthwhile to translate the first part of a linguistic term into a trapezoidal fuzzy number (TFN) using the techniques by Wang and Hao [54], such that the preference degree can be obtained in a continuous way. In the second session, reviewers are asked to evaluate the obtained preference degree TFN by voting in favor, voting against, or abstaining on each evaluation index. In this way, the final preference relations can be obtained with respect to each criterion as assigned by SVTNNs. When represented in this way, the preference information is considerably more comprehensive and accurate. Additionally, the experts complete the survey anonymously, not communicating with each other so as not to influence each other. Using this strategy, the evaluation information expressed by SVTNNs is composed of independent components on \(T\), \(I\), and \(F\). This paper only considers SVTNNs with independent inputs.
The rest of this paper is organized as follows. Section 2 reviews some preliminaries regarding SVTNNs and their operations. Section 2 also revisits existing comparison methods and analyzes their deficiencies. In order to overcome shortcomings in operations and comparison methods, Sect. 3 proposes new operations, an improved comparison method, and two aggregation operators to fuse decision information. Section 4 presents SVTNPR, exploring their complete consistency and acceptable consistency conditions. Based on these foundations, Sect. 5 presents a MCDM method using SVTNPRs. To verify the feasibility of the method, Sect. 6 provides an example of a green supplier selection problem and conducts a comparison analysis. Finally, Sect. 7 presents the main conclusions.
2 Preliminaries
This section introduces some basic concepts, operations, and comparison methods related to SVTNNs; in addition, this section briefly reviews the concepts of FPRs and IFPRs, which are utilized in the subsequent analyses.
2.1 NSs and SVTNNs
Definition 1
[55] Suppose that \(K = [a_{1} ,a_{2} ,a_{3} ,a_{4} ]\) is a TFN on the real number set R, and \(a_{1} \le a_{2} \le a_{3} \le a_{4}\). Then, its membership function \(\mu_{K} :R \to [0,1]\) is defined as follows:
When \(a_{2} = a_{3}\), the TFN \(K = [a_{1} ,a_{2} ,a_{3} ,a_{4} ]\) is reduced to a triangular fuzzy number.
Definition 2
[56] Let \(X\) be a space of points or objects, with a generic element in \(X\) denoted by \(x\). A single-valued neutrosophic set (SVNS) \(V\) in \(X\) is characterized by three independent parts, namely truth-membership function \(T_{V}\), indeterminacy-membership function \(I_{V}\), and falsity-membership function \(F_{V}\), such that \(T_{V} :X \to [0,1]\), \(I_{V} :X \to [0,1]\), and \(F_{V} :X \to [0,1]\).
For notational convenience, \(V\) is often denoted as \(V = \left\{ { < x,\left( {T_{V} (x),I_{V} (x),F_{V} (x)} \right) > |x \in X} \right\}\), satisfying \(0 \le T_{V} (x) + I_{V} (x) + F_{V} (x) \le 3\).
A SVNN, which is an element in a SVNS, is denoted by crisp numbers; it is related to a discrete set and cannot represent very much fuzzy information. In order to overcome this challenge, Ye [39] extended the discrete set to a continuous one by combining the concept of TFNs with SVNSs and defined the SVTNNs.
Definition 3
[39] Let \(T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} \in \left[ {0,1} \right]\); then, a SVTNN \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} } \right.} \right.,\left. {I_{{\tilde{a}}} ,\left. {F_{{\tilde{a}}} } \right)} \right\rangle\) is a special NS on the real number set R, whose truth-membership function \(\mu_{{\tilde{a}}}\), indeterminacy-membership function \(\nu_{{\tilde{a}}}\), and falsity-membership function \(\lambda_{{\tilde{a}}}\) are given as follows:
When \(a_{1} > 0\), \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} } \right.} \right.,\left. {I_{{\tilde{a}}} ,\left. {F_{{\tilde{a}}} } \right)} \right\rangle\) is called a positive SVTNN, denoted by \(\tilde{a} > 0\). Similarly, when \(a_{4} \le 0\), \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} } \right.} \right.,\left. {I_{{\tilde{a}}} ,\left. {F_{{\tilde{a}}} } \right)} \right\rangle\) becomes a negative SVTNN, denoted by \(\tilde{a} < 0\). When \(0 \le a_{1} \le a_{2} \le a_{3} \le a_{4} \le 1\) and \(T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} \in \left[ {0,1} \right]\), \(\tilde{a}\) is called a normalized SVTNN.
When \(I_{{\tilde{a}}} = 1 - T_{{\tilde{a}}} - F_{{\tilde{a}}}\), the SVTNN is reduced to a trapezoidal intuitionistic fuzzy number (TIFN). When \(a_{2} = a_{3}\), \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} } \right.} \right.,\left. {I_{{\tilde{a}}} ,\left. {F_{{\tilde{a}}} } \right)} \right\rangle\) turns out to be a single-valued triangular neutrosophic number (SVTrNN). When \(I_{{\tilde{a}}} = 0\), \(F_{{\tilde{a}}} = 0\), a SVTNN is reduced to a generalized TFN, \(\tilde{a} = \left\langle {[a_{1} ,a_{2} ,a_{3} ,a_{4} ],T_{{\tilde{a}}} } \right\rangle\).
2.2 Operations and comparison methods for SVTNNs
Definition 4
[39] Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {b_{1} ,b_{2} ,b_{3} ,b_{4} } \right],\left( {T_{{\tilde{b}}} ,I_{{\tilde{b}}} ,F_{{\tilde{b}}} } \right)} \right\rangle\) be two arbitrary SVTNNs, and \(\zeta \ge 0\); then, their operations are defined as follows:
-
1.
\(\tilde{a} \oplus \tilde{b} = \left\langle {\left[ {a_{1} + b_{1} ,a_{2} + b_{2} ,a_{3} + b_{3} ,a_{4} + b_{4} } \right],\left( {T_{{\tilde{a}}} + T_{{\tilde{b}}} - T_{{\tilde{a}}} T_{{\tilde{b}}} ,I_{{\tilde{a}}} I_{{\tilde{b}}} ,F_{{\tilde{a}}} F_{{\tilde{b}}} } \right)} \right\rangle\);
-
2.
\(\tilde{a} \otimes \tilde{b} = \left\langle {\left[ {a_{1} b_{1} ,a_{2} b_{2} ,a_{3} b_{3} ,a_{4} b_{4} } \right],\left( {T_{{\tilde{a}}} T_{{\tilde{b}}} ,I_{{\tilde{a}}} + I_{{\tilde{b}}} - I_{{\tilde{a}}} I_{{\tilde{b}}} ,F_{{\tilde{a}}} + F_{{\tilde{b}}} - F_{{\tilde{a}}} F_{{\tilde{b}}} } \right)} \right\rangle\);
-
3.
\(\zeta \tilde{a} = \left\langle {\left[ {\zeta a_{1} ,\zeta a_{2} ,\zeta a_{3} ,\zeta a_{4} } \right],\left( {1 - \left( {1 - T_{{\tilde{a}}} } \right)^{\zeta } ,\left( {I_{{\tilde{a}}} } \right)^{\zeta } ,\left( {F_{{\tilde{a}}} } \right)^{\zeta } } \right)} \right\rangle\); and
-
4.
\(\tilde{a}^{\zeta } = \left\langle {\left[ {a_{1}^{\zeta } ,a_{2}^{\zeta } ,a_{3}^{\zeta } ,a_{4}^{\zeta } } \right],\left( {\left( {T_{{\tilde{a}}} } \right)^{\zeta } ,1 - \left( {1 - I_{{\tilde{a}}} } \right)^{\zeta } ,1 - \left( {1 - F_{{\tilde{a}}} } \right)^{\zeta } } \right)} \right\rangle\).
However, some drawbacks exist in operations (1) and (3) in Definition 4, and they will be discussed in Example 1 and Example 2, respectively.
Example 1
Let \(\tilde{a} = \left\langle {\left[ {0.1,0.1,0.2,0.3} \right],\left( {0,0,1} \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {0.1,0.1,0.2,0.3} \right],\left( {1,0,0} \right)} \right\rangle\) be two SVTNNs. According to Definition 4, the following result can be calculated: \(\tilde{a} + \tilde{b} = \left\langle {\left[ {0.2,0.2,0.4,0.6} \right],\left( {1,0,0} \right)} \right\rangle\). However, this result is inaccurate because it does not consider the falsity-membership of \(\tilde{a}\), the correlations among TFNs and the membership degrees of \(\tilde{a}\) and \(\tilde{b}\). Therefore, these operations are imprecise.
Example 2
Let \(\tilde{a}_{1} = \left\langle {[0.03,0.05,0.07,0.09],(0.3,0.5,0.5)} \right\rangle\) be a SVTNN and \(\zeta = 10\). Then, the result \(\zeta \tilde{a}_{1}\) obtained using Definition 4 is
According to Example 2, the former TFN and the latter SVNN in \(\tilde{a}_{1}\) are calculated simultaneously; therefore, repeated calculations occur between the three membership degrees and the TFN of the SVTNN, which significantly distort the result and conflict with common sense.
Therefore, some new operations for SVTNNs must be defined in order to overcome these anomalies. The new operations are discussed in Sect. 3.
To compare any two SVTNNs, Ye [39] and Deli and Şubaş [57] proposed comparison methods based on the score function and accuracy function.
Definition 5
[39] Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) be a SVTNN. The score function of \(\tilde{a}\) is defined as
Definition 6
[39] Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {b_{1} ,b_{2} ,b_{3} ,b_{4} } \right],\left( {T_{{\tilde{b}}} ,I_{{\tilde{b}}} ,F_{{\tilde{b}}} } \right)} \right\rangle\) be two SVTNNs. Then,
-
1.
when \(S\left( {\tilde{a}} \right) > S\left( {\tilde{b}} \right)\), \(\tilde{a} > \tilde{b}\); and
-
2.
when \(S\left( {\tilde{a}} \right) = S\left( {\tilde{b}} \right)\), \(\tilde{a} = \tilde{b}\).
However, some flaws exist in Definition 5, which are discussed in Example 3.
Example 3
Let \(\tilde{a} = \left\langle {\left[ {0.3,0.4,0.5,0.8} \right],\left( {0.5,0.3,0.7} \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {0.5,0.7,0.8,1} \right],\left( {0.2,0.8,0.4} \right)} \right\rangle\) be two SVTNNs; then, it is clear that \(\tilde{a} \ne \tilde{b}\). However, according to Definitions 5 and 6, \(S\left( {\tilde{a}} \right) = S\left( {\tilde{b}} \right) = 0.25\), and \(\tilde{a} = \tilde{b}\), which does not conform to our intuition.
In order to overcome the deficiency in this comparison method, Deli and Şubaş [57] defined a new comparison method for SVTNNs.
Definition 7
[57] Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) be a SVTNN. The score function and accuracy function of \(\tilde{a}\) are defined, respectively, as follows:
Definition 8
[57] Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {b_{1} ,b_{2} ,b_{3} ,b_{4} } \right],\left. {\left( {T_{{\tilde{b}}} ,I_{{\tilde{b}}} ,F_{{\tilde{b}}} } \right)} \right\rangle } \right.\) be two arbitrary SVTNNs.
-
1.
When \(S^{\prime}\left( {\tilde{a}} \right) < S^{\prime}\left( {\tilde{b}} \right)\), \(\tilde{a} < \tilde{b}\);
-
2.
when \(S^{\prime}\left( {\tilde{a}} \right) = S^{\prime}\left( {\tilde{b}} \right)\) and \(H^{\prime}\left( {\tilde{a}} \right) < H^{\prime}\left( {\tilde{b}} \right)\), \(\tilde{a} < \tilde{b}\); and
-
3.
when \(S^{\prime}\left( {\tilde{a}} \right) = S^{\prime}\left( {\tilde{b}} \right)\) and \(H^{\prime}\left( {\tilde{a}} \right) = H^{\prime}\left( {\tilde{b}} \right)\), \(\tilde{a} = \tilde{b}\).
Although the comparison method proposed in Deli and Şubaş [57] addressed the problem in Ye [39], some drawbacks still exist in the operations, as described in Example 4.
Example 4
Let \(\tilde{a} = \left\langle {\left[ {0.2,0.3,0.5,0.8} \right],\left( {0.1,0.8,0} \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {0.1,0.4,0.5,0.8} \right],\left( {0.2,0.9,0} \right)} \right\rangle\) be two SVTNNs; then, it is clear that \(\tilde{a} \ne \tilde{b}\). According to Definition 7, we can determine that \(S^{\prime}\left( {\tilde{a}} \right) = S^{\prime}\left( {\tilde{b}} \right) = 0.146\), \(H^{\prime}\left( {\tilde{a}} \right) = H^{\prime}\left( {\tilde{b}} \right) = 0.146\), and according to Definition 8, \(\tilde{a} = \tilde{b}\), which is counterintuitive.
As a result, it is worthwhile to define a new comparison method for SVTNNs that overcomes the shortcomings of the extant research.
2.3 Fuzzy preference relation and intuitionistic fuzzy preference relation
Definition 9
[53] A FPR \(B\) on the set \(A = \left\{ {A_{1} ,A_{2} , \ldots ,A_{m} } \right\}\) is represented by a matrix \(B = \left( {b_{ik} } \right)_{m \times m}\), where \(b_{ik}\) is the intensity of preference of \(A_{i}\) over \(A_{k}\), and satisfies
In the FPRs, the preference degree is represented by a crisp number. However, this seems to be counterintuitive for several reasons: (1) The experiences and knowledge of experts are limited and they may be not familiar with the content of the decision-making problems; (2) even if the experts are familiar with the decision-making problems, scarcity of information and time pressure may make it difficult for DMs to determine the exact values of the preference values; and (3) the evaluation information for alternatives usually contain some incomplete, inconsistent, or indeterminate types. All of the situations described above can create challenges when experts attempt to construct a FPR when comparing alternatives.
In order to express both the vagueness and uncertainty existing in DMs’ pairwise judgments, Xu [52] proposed a standard definition of an IFPR, in which experts can express their opinions of the alternatives from three perspectives: preferred, non-preferred, and indeterminate.
Definition 10
[52] An IFPR \(R\) on the set \(X = \left\{ {x_{1} ,x_{2} , \ldots ,x_{m} } \right\}\) is represented by a matrix \(R = \left( {r_{ij} } \right)_{m \times m}\), where \(r_{ij} = \left\langle {\left( {x_{i} ,x_{j} } \right),\left( {u\left( {x_{i} ,x_{j} } \right),v\left( {x_{i} ,x_{j} } \right),\pi \left( {x_{i} ,x_{j} } \right)} \right)} \right\rangle\) for all \(i,j = 1,2, \ldots ,m\). For simplicity, \(r_{ij}\) is denoted as \(r_{ij} = \left( {u_{ij} ,v_{ij} ,\pi_{ij} } \right)\), and \(u_{ij} + v_{ij} + \pi_{ij} = 1\), \(u_{ij}\), \(v_{ij} \in \left[ {0,1} \right]\), for all \(i,j = 1,2, \ldots ,m\), satisfying
-
1.
\(u_{ij} + v_{ij} \le 1\);
-
2.
\(u_{ij} = v_{ji}\) and \(v_{ij} = u_{ji}\); and
-
3.
\(u_{ii} = v_{ii} = 0.5\);
where \(u_{ij}\) indicates the certainty degree to which alternative \(X_{i}\) is preferred to \(X_{j}\), \(v_{ij}\) denotes the certainty degree to which the alternative \(X_{i}\) is non-preferred to \(X_{j}\), and \(\pi_{ij}\) is interpreted as the indeterminate degree to which the alternative \(X_{i}\) is superior to \(X_{j}\).
However, part of the information about the alternatives in complex decision-making problems, including inconsistent or unknown decision information, still cannot be depicted in depth. For this reason, it is worthwhile to extend IFPRs to SVTNPRs; this extension is discussed in Sect. 4.
3 New operations and comparison method for SVTNNs
This section defines new operations and proves their properties. Moreover, a new comparison method is proposed on the basis of score and accuracy functions in order to cover the limitations presented in Sect. 2.2. Finally, two aggregation operators are proposed.
3.1 New operations based on area of SVTNNs
Definition 11
Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) be an arbitrary SVTNN; then, the areas under the three membership functions, denoted, respectively, by \({\text{ar}}\left( {T_{{\tilde{a}}} } \right)\), \({\text{ar}}\left( {I_{{\tilde{a}}} } \right)\) and \({\text{ar}}\left( {F_{{\tilde{a}}} } \right)\), can be defined as follows:
Definition 12
Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {b_{1} ,b_{2} ,b_{3} ,b_{4} } \right],\left( {T_{{\tilde{b}}} ,I_{{\tilde{b}}} ,F_{{\tilde{b}}} } \right)} \right\rangle\) be two arbitrary SVTNNs, and \(\zeta \ge 0\); then, new operations for SVTNNs are defined as follows:
1. \(\tilde{a} \oplus \tilde{b} = \left\langle {\left[ {a_{1} + b_{1} ,a_{2} + b_{2} ,a_{3} + b_{3} ,a_{4} + b_{4} } \right],\left( {\frac{{\varphi (\tilde{a})T_{{\tilde{a}}} + \varphi (\tilde{b})T_{{\tilde{b}}} }}{{\varphi (\tilde{a}) + \varphi (\tilde{b})}},} \right.} \right.\frac{{\varphi (\tilde{a})\left( {1 - I_{{\tilde{a}}} } \right) + \varphi (\tilde{b})\left( {1 - I_{{\tilde{b}}} } \right)}}{{\varphi (\tilde{a}) + \varphi (\tilde{b})}},\)
\(\left. {\left. {\frac{{\varphi (\tilde{a})\left( {1 - F_{{\tilde{a}}} } \right) + \varphi (\tilde{b})\left( {1 - F_{{\tilde{b}}} } \right)}}{{\varphi (\tilde{a}) + \varphi (\tilde{b})}}} \right)} \right\rangle\), where \(\varphi (\tilde{a}) = \frac{{a_{3} - a_{2} + a_{4} - a_{1} }}{2}\) and \(\varphi (\tilde{b}) = \frac{{b_{3} - b_{2} + b_{4} - b_{1} }}{2}\);
2. \(\tilde{a} - \tilde{b} = \left\langle {\left[ {a_{1} - b_{4} ,a_{2} - b_{3} ,a_{3} - b_{2} ,a_{4} + b_{1} } \right],\left( {\frac{{\varphi (\tilde{a})T_{{\tilde{a}}} + \varphi (\tilde{b})T_{{\tilde{b}}} }}{{\varphi (\tilde{a}) + \varphi (\tilde{b})}},} \right.} \right.\frac{{\varphi (\tilde{a})\left( {1 - I_{{\tilde{a}}} } \right) + \varphi (\tilde{b})\left( {1 - I_{{\tilde{b}}} } \right)}}{{\varphi (\tilde{a}) + \varphi (\tilde{b})}},\)
\(\left. {\left. {\frac{{\varphi (\tilde{a})\left( {1 - F_{{\tilde{a}}} } \right) + \varphi (\tilde{b})\left( {1 - F_{{\tilde{b}}} } \right)}}{{\varphi (\tilde{a}) + \varphi (\tilde{b})}}} \right)} \right\rangle\), where \(\varphi (\tilde{a}) = \frac{{a_{3} - a_{2} + a_{4} - a_{1} }}{2}\) and \(\varphi (\tilde{b}) = \frac{{b_{3} - b_{2} + b_{4} - b_{1} }}{2}\);
Note When \(\varphi (\tilde{a}) = 0\) and \(\varphi \left( {\tilde{b}} \right) = 0\), then \(T_{{\tilde{a} \oplus \tilde{b}}} = T_{{\tilde{a} - \tilde{b}}} = \frac{{T_{{\tilde{a}}} + T_{{\tilde{b}}} }}{2}\), \(I_{{\tilde{a} \oplus \tilde{b}}} = I_{{\tilde{a} - \tilde{b}}} = \frac{{I_{{\tilde{a}}} + I_{{\tilde{b}}} }}{2}\), and \(F_{{\tilde{a} \oplus \tilde{b}}} = F_{{\tilde{a} - \tilde{b}}} = \frac{{F_{{\tilde{a}}} + F_{{\tilde{b}}} }}{2}\).
3. \(\tilde{a} \otimes \tilde{b} = \left\langle {\left[ {a_{1} b_{1} ,a_{2} b_{2} ,a_{3} b_{3} ,a_{4} b_{4} } \right],\left( {T_{{\tilde{a}}} T_{{\tilde{b}}} ,I_{{\tilde{a}}} + I_{{\tilde{b}}} - I_{{\tilde{a}}} I_{{\tilde{b}}} ,F_{{\tilde{a}}} + F_{{\tilde{b}}} - F_{{\tilde{a}}} F_{{\tilde{b}}} } \right)} \right\rangle\);
4. \(\zeta \tilde{a} = \left\langle {\left[ {\zeta a_{1} ,\zeta a_{2} ,\zeta a_{3} ,\zeta a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\), \(\zeta \ge 0\);
5. \(\tilde{a}^{\zeta } = \left\langle {\left[ {a_{1}^{\zeta } ,a_{2}^{\zeta } ,a_{3}^{\zeta } ,a_{4}^{\zeta } } \right],\left( {\left( {T_{{\tilde{a}}} } \right)^{\zeta } ,1 - \left( {1 - I_{{\tilde{a}}} } \right)^{\zeta } ,1 - \left( {1 - F_{{\tilde{a}}} } \right)^{\zeta } } \right)} \right\rangle\), \(\zeta \ge 0\);
6. \({\text{neg}}\left( {\tilde{a}} \right) = \left\langle {{\text{neg}}\left( K \right),\left( {F_{{\tilde{a}}} ,1 - I_{{\tilde{a}}} ,T_{{\tilde{a}}} } \right)} \right\rangle\), where \(K\) denotes the TFN in \(\tilde{a}\).
Example 5
Using the data in Example 1, let \(\zeta = 2\); based on Definition 12, we can identify that
-
1.
\(\tilde{a} \oplus \tilde{b}{ = }\left\langle {\left[ {0.2,0.2,0.35,0.7} \right],\left( {0.538,0,0.538} \right)} \right\rangle\);
-
2.
\(\tilde{a} \otimes \tilde{b}{ = }\left\langle {\left[ {0.01,0.01,0.03,0.12} \right],\left( {0,0,0} \right)} \right\rangle\);
-
3.
\(2\tilde{a}{ = }\left\langle {\left[ {0.2,0.2,0.4,0.6} \right],\left( {0,0,1} \right)} \right\rangle\); and
-
4.
\(\tilde{a}^{2} { = }\left\langle {\left[ {0.01,0.01,0.04,0.09} \right],\left( {0,0,1} \right)} \right\rangle\).
Compared with the operations proposed by Ye [39], our new proposed operations for SVTNNs not only capture the correlations of TFNs and the three membership degrees of SVTNNs, but also effectively avoid the loss and distortion of information. Therefore, the newly defined operations are more reasonable than the existing ones.
In terms of the corresponding operations for SVTNNs, the following theorem can be easily proved.
Theorem 1
Let \(\tilde{a}\), \(\tilde{b}\) and \(\tilde{c}\) be three SVTNNs and \(\zeta \ge 0\); then, the following equations are true:
-
1.
\(\tilde{a} \oplus \tilde{b} = \tilde{b} \oplus \tilde{a}\);
-
2.
\(\left( {\tilde{a} \oplus \tilde{b}} \right) \oplus \tilde{c}{ = }\tilde{a} \oplus \left( {\tilde{b} \oplus \tilde{c}} \right)\);
-
3.
\(\tilde{a} \otimes \tilde{b} = \tilde{b} \otimes \tilde{a}\);
-
4.
\(\left( {\tilde{a} \otimes \tilde{b}} \right) \otimes \tilde{c}{ = }\tilde{a} \otimes \left( {\tilde{b} \otimes \tilde{c}} \right)\);
-
5.
\(\zeta \tilde{a} \oplus \zeta \tilde{b} = \zeta \left( {\tilde{b} \oplus \tilde{a}} \right)\); and
-
6.
\(\left( {\tilde{a} \otimes \tilde{b}} \right)^{\zeta } { = }\tilde{a}^{\zeta } \otimes \tilde{b}^{\zeta }\).
The proof of Theorem 1 according to Definition 12 is self-explanatory, so it is omitted here.
3.2 A new comparison method for SVTNNs
Motivated by the centroid defuzzification method [54, 58] and related research on neutrosophic theory [35, 59], we redefine the score function, accuracy function, and certainty function of SVTNNs, based on which a new comparison method for SVTNNs is presented.
Definition 13
[54] Let \(K = \left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right]\) be a TFN on the real number set R, and \(a_{1} \le a_{2} \le a_{3} \le a_{4}\); then, the center of gravity (COG) of \(K\) can be defined as follows:
Definition 14
Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) be a SVTNN; then, the score function, accuracy function, and certainty function of SVTNN \(\tilde{a}\) are defined, respectively, as:
Based on the above three functions, we can define a novel comparison method for SVTNNs as follows.
Definition 15
Let \(\tilde{a} = \left\langle {\left[ {a_{1} ,a_{2} ,a_{3} ,a_{4} } \right],\left( {T_{{\tilde{a}}} ,I_{{\tilde{a}}} ,F_{{\tilde{a}}} } \right)} \right\rangle\) and \(\tilde{b} = \left\langle {\left[ {b_{1} ,b_{2} ,b_{3} ,b_{4} } \right],\left( {T_{{\tilde{b}}} ,I_{{\tilde{b}}} ,F_{{\tilde{b}}} } \right)} \right\rangle\) be two SVTNNs. The comparison method for \(\tilde{a}\) and \(\tilde{b}\) can be defined as follows:
-
1.
When \(E\left( {\tilde{a}} \right) > E\left( {\tilde{b}} \right)\), \(\tilde{a} > \tilde{b}\), meaning that \(\tilde{a}\) is superior to \(\tilde{b}\);
-
2.
when \(E\left( {\tilde{a}} \right) = E\left( {\tilde{b}} \right)\), and \(A\left( {\tilde{a}} \right) > A\left( {\tilde{b}} \right)\), \(\tilde{a} > \tilde{b}\), meaning that \(\tilde{a}\) is superior to \(\tilde{b}\);
-
3.
when \(E\left( {\tilde{a}} \right) = E\left( {\tilde{b}} \right)\), and \(A\left( {\tilde{a}} \right) < A\left( {\tilde{b}} \right)\), \(\tilde{a} < \tilde{b}\), meaning that \(\tilde{a}\) is inferior to \(\tilde{b}\); and
-
4.
when \(E\left( {\tilde{a}} \right) = E\left( {\tilde{b}} \right)\), \(A\left( {\tilde{a}} \right) = A\left( {\tilde{b}} \right)\), and \(C\left( {\tilde{a}} \right) > C\left( {\tilde{b}} \right)\), \(\tilde{a} > \tilde{b}\), meaning that \(\tilde{a}\) is superior to \(\tilde{b}\); and \(\tilde{a} < \tilde{b}\) when \(C\left( {\tilde{a}} \right) < C\left( {\tilde{b}} \right)\), meaning that \(\tilde{a}\) is inferior to \(\tilde{b}\); and \(\tilde{a} = \tilde{b}\) when \(C\left( {\tilde{a}} \right){ = }C\left( {\tilde{b}} \right)\), meaning that \(\tilde{a}\) is indifferent to \(\tilde{b}\).
Example 6
Utilizing the data in Example 1, we can calculate that \(E\left( {\tilde{a}} \right) = 0.1986\) and \(E\left( {\tilde{b}} \right) = 0.195\). Then, \(\tilde{a} > \tilde{b}\); that is, \(\tilde{b}\) is superior to \(\tilde{a}\), which is consistent with our intuition.
3.3 Aggregation operators for SVTNNs
Definition 16
Let \(\tilde{a}_{j} = \left\langle {\left[ {a_{j1} ,a_{j2} ,a_{j3} ,a_{j4} } \right],\left( {T_{{\tilde{a}_{j} }} ,I_{{\tilde{a}_{j} }} ,F_{{\tilde{a}_{j} }} } \right)} \right\rangle\) \(\left( {j = 1,2, \ldots ,n} \right)\) be a set of SVTNNs, and let \(\varpi = \left( {\varpi_{1} ,\varpi_{2} , \ldots ,\varpi_{n} } \right)^{T}\) be the weight vector of \(\tilde{a}_{j}\) \(\left( {j = 1,2, \ldots ,n} \right)\) with \(\varpi_{j} \in \left[ {0,1} \right]\) and \(\sum\nolimits_{j = 1}^{n} {\varpi_{j} } = 1\); then, the SVTNWAA operator can be defined as follows:
Theorem 2
Let \(\tilde{a}_{j} = \left\langle {\left[ {a_{j1} ,a_{j2} ,a_{j3} ,a_{j4} } \right],\left( {T_{{\tilde{a}_{j} }} ,I_{{\tilde{a}_{j} }} ,F_{{\tilde{a}_{j} }} } \right)} \right\rangle\) \(\left( {j = 1,2, \ldots ,n} \right)\) be a set of SVTNNs. Then, the aggregated value utilizing the SVTNWAA operator is still a SVTNN, which is shown as follows:
where \(\varphi (\tilde{a}_{j} ) = \frac{{a_{j3} - a_{j2} + a_{j4} - a_{j1} }}{2}\).
In the following, we proof Theorem 2 using mathematical induction on \(n\).
Proof
1. When \(n = 2\), the following equation can be obtained:
where \(\varphi (\tilde{a}_{1} ) = \frac{{a_{13} - a_{12} + a_{14} - a_{11} }}{2}\) and \(\varphi (\tilde{a}_{2} ) = \frac{{a_{23} - a_{22} + a_{24} - a_{21} }}{2}\). Clearly, when \(n = 2\), Theorem 2 is true.
2. Suppose that when \(n = k\), Theorem 2 is true. That is,
Then, when \(n = k{ + }1\), the following equation can be calculated:
That is, Theorem 2 is true for \(n = k{ + }1\).
Therefore, Theorem 2 holds for all \(n\).
Definition 17
Let \(\tilde{a}_{j} = \left\langle {\left[ {a_{j1} ,a_{j2} ,a_{j3} ,a_{j4} } \right],\left( {T_{{\tilde{a}_{j} }} ,I_{{\tilde{a}_{j} }} ,F_{{\tilde{a}_{j} }} } \right)} \right\rangle\) \(\left( {j = 1,2, \ldots ,n} \right)\) be a set of SVTNNs, and \(\varpi = \left( {\varpi_{1} ,\varpi_{2} , \ldots ,\varpi_{n} } \right)^{T}\) be the weight vector of \(\tilde{a}_{j}\) \(\left( {j = 1,2, \ldots ,n} \right)\) with \(\varpi_{j} \in \left[ {0,1} \right]\) and \(\sum\nolimits_{j = 1}^{n} {\varpi_{j} } = 1\); then, the SVTNWGA operator is defined as follows:
Theorem 3
Let \(\tilde{a}_{j} = \left\langle {\left[ {a_{j1} ,a_{j2} ,a_{j3} ,a_{j4} } \right],\left( {T_{{\tilde{a}_{j} }} ,I_{{\tilde{a}_{j} }} ,F_{{\tilde{a}_{j} }} } \right)} \right\rangle\) \(\left( {j = 1,2, \ldots ,n} \right)\) be a set of SVTNNs. Then, the aggregated value utilizing the SVTNWGA operator is still a SVTNN. The aggregated result satisfies
The proof is similar to Theorem 2, so it is omitted here.
4 Single-valued trapezoidal neutrosophic preference relations
This section extends the traditional FPRs and IFPRs to SVTNPRs and explores some of their prominent characteristics.
Definition 18
Let a matrix \(\widetilde{R} = [\tilde{a}_{ij} ]_{m \times m}\) on the set \(A = \left\{ {A_{1} ,A_{2} , \ldots ,A_{m} } \right\}\) be composed of SVTNNs, where \(\tilde{a}_{ij} = \left\langle {\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right],\left( {T_{{\tilde{a}_{ij} }} ,I_{{\tilde{a}_{ij} }} ,F_{{\tilde{a}_{ij} }} } \right)} \right\rangle\) for all \(i,j = 1,2, \ldots ,m\), and \(0 \le a_{ij}^{1} \le a_{ij}^{2} \le a_{ij}^{3} \le a_{ij}^{4} \le 1\), \(0 \le T_{{\tilde{a}_{ij} }}\), \(I_{{\tilde{a}_{ij} }}\), \(F_{{\tilde{a}_{ij} }} \le 1\). When comparing alternative \(A_{i}\) over \(A_{j}\), \(T_{{\tilde{a}_{ij} }}\) indicates the certainty degree to which alternative \(A_{i}\) is preferred to \(A_{j}\) with a degree of \(\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right]\) which is represented as a TFN; \(I_{{\tilde{a}_{ij} }}\) is the indeterminate degree to which alternative \(A_{i}\) is preferred to \(A_{j}\) with a degree of \(\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right]\); and \(F_{{\tilde{a}_{ij} }}\) denotes the certainty degree to which alternative \(A_{i}\) is non-preferred to \(A_{j}\) with a degree of \(\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right]\).
The matrix \(\widetilde{R} = [\tilde{a}_{ij} ]_{m \times m}\) is denoted as a SVTNPR on \(A\), if the following rules can be satisfied:
-
1.
\(a_{ij}^{1} + a_{ji}^{4} = 1\), \(a_{ij}^{2} + a_{ji}^{3} = 1\), \(a_{ij}^{3} + a_{ji}^{2} = 1\), \(a_{ij}^{4} + a_{ji}^{1} = 1\);
-
2.
\(T_{{\tilde{a}_{ij} }} = F_{{\tilde{a}_{ji} }}\), \(I_{{\tilde{a}_{ij} }} = I_{{\tilde{a}_{ji} }}\), \(F_{{\tilde{a}_{ij} }} = T_{{\tilde{a}_{ji} }}\);
-
3.
\(\tilde{a}_{ii} = \left\langle {\left[ {1,1,1,1} \right],\left( {1,0,0} \right)} \right\rangle\); and
-
4.
\(0 \le T_{{\tilde{a}_{ij} }} + I_{{\tilde{a}_{ij} }} + F_{{\tilde{a}_{ij} }} \le 3\).
Theorem 4
A SVTNPR \(\widetilde{R}\) on the set of \(A = \left\{ {A_{1} ,A_{2} , \ldots ,A_{m} } \right\}\) is represented by a matrix \(\widetilde{R} = [\tilde{a}_{ij} ]_{m \times m}\), where \(\tilde{a}_{ij} = \left\langle {\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right],\left( {T_{{\tilde{a}_{ij} }} ,I_{{\tilde{a}_{ij} }} ,F_{{\tilde{a}_{ij} }} } \right)} \right\rangle\) for all \(i,j = 1,2, \ldots ,m\), when the elements in the ith row and ith column are removed from \(\widetilde{R} = [\tilde{a}_{ij} ]_{m \times m}\), the preference relation composed by the remainder elements of \(\widetilde{R}\) is still a SVTNPR.
Proof
The proof of Theorem 4 can be completed easily according to Definition 18, so it is omitted here.
Definition 19
A SVTNPR \(\widetilde{R} = [\tilde{a}_{ij} ]_{m \times m}\) with \(\tilde{a}_{ij} = \left\langle {\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right],\left( {T_{{\tilde{a}_{ij} }} ,I_{{\tilde{a}_{ij} }} ,F_{{\tilde{a}_{ij} }} } \right)} \right\rangle\) \(\left( {i,j,k = 1,2, \ldots ,m} \right)\) is called a completely consistent SVTNPR when the following statements are equivalent for any \(i,j,k\):
-
1.
\(a_{ij}^{1} + a_{ij}^{4} + a_{jk}^{1} + a_{jk}^{4} + a_{ki}^{1} + a_{ki}^{4} = a_{ij}^{2} + a_{ij}^{3} + a_{jk}^{2} + a_{jk}^{3} + a_{ki}^{2} + a_{ki}^{3} = 3\);
-
2.
\(T_{{\tilde{a}_{ij} }} T_{{\tilde{a}_{jk} }} T_{{\tilde{a}_{ki} }} = F_{{\tilde{a}_{ji} }} F_{{\tilde{a}_{ik} }} F_{{\tilde{a}_{kj} }}\);
-
3.
\(I_{{\tilde{a}_{ij} }} I_{{\tilde{a}_{jk} }} I_{{\tilde{a}_{ki} }} = I_{{\tilde{a}_{ji} }} I_{{\tilde{a}_{ik} }} I_{{\tilde{a}_{kj} }}\); and
-
4.
\(F_{{\tilde{a}_{ij} }} F_{{\tilde{a}_{jk} }} F_{{\tilde{a}_{ik} }} = T_{{\tilde{a}_{ji} }} T_{{\tilde{a}_{ki} }} T_{{\tilde{a}_{kj} }}\).
Let the symbols \(\succ\) and \(\sim\) be two binary relations on SVTNRs, interpreted as preferred and indifferent relations, respectively.
Definition 20
ASVTNPR \(\widetilde{R} = [\tilde{a}_{ij} ]_{m \times m}\) with \(\tilde{a}_{ij} = \left\langle {\left[ {a_{ij}^{1} ,a_{ij}^{2} ,a_{ij}^{3} ,a_{ij}^{4} } \right],\left( {T_{{\tilde{a}_{ij} }} ,I_{{\tilde{a}_{ij} }} ,F_{{\tilde{a}_{ij} }} } \right)} \right\rangle\) \(\left( {i,j,k = 1,2, \ldots ,m} \right)\) is called an acceptably consistent SVTNPR when the following statements are equivalent for any \(i,j,k\):
-
1.
when \(\tilde{a}_{ij} \succ \tilde{a}_{ii}\) and \(\tilde{a}_{jk} \succ \tilde{a}_{ii}\), then \(\tilde{a}_{ik} \succ \tilde{a}_{ii}\);
-
2.
when \(\tilde{a}_{ij} \sim \tilde{a}_{ii}\) and \(\tilde{a}_{jk} \sim \tilde{a}_{ii}\), then \(\tilde{a}_{ik} \sim \tilde{a}_{ii}\); and
-
3.
when \(\tilde{a}_{ij} \prec \tilde{a}_{ii}\) and \(\tilde{a}_{jk} \prec \tilde{a}_{ii}\), then \(\tilde{a}_{ik} \prec \tilde{a}_{ii}\).
It can be interpreted as follows: when \(\tilde{a}_{ij}\) is preferred (inferior) to \(\tilde{a}_{ii}\), and \(\tilde{a}_{jk}\) is preferred (inferior) to \(\tilde{a}_{ii}\), \(\tilde{a}_{ij}\) should be preferred (inferior) to \(\tilde{a}_{jk}\); similarly, when \(\tilde{a}_{ij}\) is indifferent to \(\tilde{a}_{ii}\), and \(\tilde{a}_{jk}\) is indifferent to \(\tilde{a}_{ii}\), \(\tilde{a}_{ij}\) should be indifferent to \(\tilde{a}_{jk}\), too.
Clearly, when \(a_{2} = a_{3}\), \(T_{{\tilde{a}_{ij} }} = 1\), and \(I_{{\tilde{a}_{ij} }} = F_{{\tilde{a}_{ij} }} = 0\), the SVTNPR is reduced to a triangular fuzzy preference relation (TrFPR); meanwhile, when \(a_{2} = a_{3}\) and \(F_{{\tilde{a}_{ij} }} = 1 - T_{{\tilde{a}_{ij} }} - I_{{\tilde{a}_{ij} }}\), the SVTNPR is reduced to a triangular intuitionistic fuzzy preference relation (TrIPR). In essence, a SVTNPR is a generalized form of FPR.
5 Multi-criteria decision-making method based on SVTNPRs and complete weight information
This section proposes an MCDM approach based on SVTNPRs and provides the main procedures of the proposed method.
For an MCDM problem with a finite set of m alternatives, let \(A = \left\{ {A_{1} ,A_{2} , \ldots ,A_{m} } \right\}\) be the set of \(m\) feasible alternatives, and let \(C = \left\{ {C_{1} ,C_{2} , \ldots ,C_{n} } \right\}\) be the set of criteria. Assume that the criteria weight vector is \(\varpi = \left( {\varpi_{1} ,\varpi_{2} , \ldots ,\varpi_{n} } \right)^{\text{T}}\) with \(\varpi_{j} \in \left[ {0,1} \right]\) and \(\sum\nolimits_{j = 1}^{n} {\varpi_{j} } = 1\). The pairwise comparison analyses are conducted with respect to every criterion \(C_{j} \left( {j = 1,2, \ldots ,n} \right)\); then, the evaluation values represented by SVTNPRs \(R_{j} = [\tilde{a}_{ik}^{j} ]_{m \times m}\) with \(\tilde{a}_{ik}^{j} = \left\langle {\left[ {a_{ik}^{1j} ,a_{ik}^{2j} ,a_{ik}^{3j} ,a_{ik}^{4j} } \right],\left( {T_{{\tilde{a}_{ik}^{j} }} ,I_{{\tilde{a}_{ik}^{j} }} ,F_{{\tilde{a}_{ik}^{j} }} } \right)} \right\rangle\) \(\left( {i = 1,2, \ldots ,m;k = 1,2, \ldots ,m;j = 1,2, \ldots ,n} \right)\) can be obtained by transforming the preference values provided by DMs.
Then, we can derive the decision matrix \(R_{j} = [\tilde{a}_{ik}^{j} ]_{m \times m}\) as follows:
Our proposed approach involves the following steps:
-
Step 1 Normalize the SVTNPRs.
Normalize the decision-making information \(\tilde{a}_{ik}^{j}\) in the matrices \(R_{j} = [\tilde{a}_{ik}^{j} ]_{m \times m}\). The criteria in the decision matrices are classified as benefit and cost types. In order to make the criterion type uniform, the cost criteria must be transformed into benefits using the negation operation defined in Definition 12. Because the elements in the TFN of SVTNNs are assigned values between 0 and 1, we conduct the negation operation as follows:
The normalized evaluation information matrices are denoted as \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m}\).
-
Step 2 Check the complete consistency or acceptable consistency of SVTNPR \(\bar{R}_{j}\).
If \(\bar{R}_{j}\) meets complete consistency or acceptable consistency, then proceed to Step 6. Otherwise, \(\bar{R}_{j}\) should be modified based on Definitions 19 and 20, until it has complete consistency.
-
Step 3 Improve the consistency degree of SVTNPR \(\bar{R}_{j}\).
Determine the element with highest degree of confidence level when comparing with other alternatives in SVTNPR \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m}\). Assume that the experts are sure about the evaluation information of the first row in \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m}\); then, according to Definition 18, the experts are also sure about elements of the first column in \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m}\).
Let the elements of the TFNs in the \(i\)th row be divided by elements in the first row. If the obtained difference is a fixed value, and the subsequent three membership degrees meet the conditions in Definition 19, then there is no need to modify; otherwise, the elements that do not meet the conditions must be modified according to Definition 19.
-
Step 4 Obtain the completely consistent SVTNPR.
Check every element in SVTNPR \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m}\) and modify the elements that do not meet the conditions until \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m}\) is completely consistent.
-
Step 5 Obtain the overall preference information.
Utilizing Eqs. (13) and (16), the overall value of alternative \(A_{i}\) can be aggregated. We can obtain the overall preference degrees of \(A_{i}\) \(\left( {i = 1,2, \ldots ,m} \right)\) when comparing it with other alternatives, and they are denoted by \(t_{ik}\) and \(t^{\prime}_{ik}\) \(\left( {i = 1,2, \ldots ,m;k = 1,2, \ldots ,m} \right)\), respectively; then, the matrices are denoted as \(T = \left( {t_{ik} } \right)_{m \times m}\) and \(T^{\prime} = \left( {t^{\prime}_{ik} } \right)_{m \times m}\), respectively.
-
Step 6 Calculate the ordering vector of each alternative.
Aggregate each row of the SVTNPRs \(T = \left( {t_{ik} } \right)_{m \times m}\) and \(T^{\prime} = \left( {t^{\prime}_{ik} } \right)_{m \times m}\), and gain the matrices \(U = \left( {u_{i} } \right)_{m \times 1}\) and \(U^{\prime} = \left( {u^{\prime}_{i} } \right)_{m \times 1}\), which are composed of the ordering vector of each alternative.
Then, the elements \(u_{i} = \left\langle {\left[ {u_{i}^{1} ,u_{i}^{2} ,u_{i}^{3} ,u_{i}^{4} } \right],\left( {T_{{u_{i} }} ,I_{{u_{i} }} ,F_{{u_{i} }} } \right)} \right\rangle\) in \(U\) should be normalized to \(\tilde{u}_{i} = \left\langle {\left[ {\tilde{u}_{i}^{1} ,\tilde{u}_{i}^{2} ,\tilde{u}_{i}^{3} ,\tilde{u}_{i}^{4} } \right],\left( {T_{{\tilde{u}_{i} }} ,I_{{\tilde{u}_{i} }} ,} \right.} \right.\left. {\left. {F_{{\tilde{u}_{i} }} } \right)} \right\rangle\) in \(V = \left( {\tilde{u}_{i} } \right)_{m \times 1}^{\text{T}} = \left( {\tilde{u}_{1} ,\tilde{u}_{2} , \ldots ,\tilde{u}_{m} } \right)\), where
Similarly, the elements \(u^{\prime}_{i} = \left\langle {\left[ {u_{i}^{{{\prime }1}} ,u_{i}^{{{\prime }2}} ,u_{i}^{{{\prime }3}} ,u_{i}^{{{\prime }4}} } \right],\left( {T_{{u^{\prime}}} ,I_{{u^{\prime}}} ,F_{{u^{\prime}}} } \right)} \right\rangle\) in \(U^{\prime}\) should be normalized to \(u^{\prime\prime}_{i} = \left\langle {\left[ {u_{i}^{{{\prime \prime }1}} ,u_{i}^{{{\prime \prime }2}} ,u_{i}^{{{\prime \prime }3}} ,u_{i}^{{{\prime \prime }4}} } \right],\left( {T_{{u^{\prime\prime}_{i} }} ,I_{{u^{\prime\prime}_{i} }} ,} \right.} \right.\) \(\left. {\left. {F_{{u^{\prime\prime}_{i} }} } \right)} \right\rangle\) in \(V^{\prime} = \left( {u^{\prime\prime}_{i} } \right)_{m \times 1}^{\text{T}} = \left( {u^{\prime\prime}_{1} ,u^{\prime\prime}_{2} , \ldots ,u^{\prime\prime}_{m} } \right)\), where
-
Step 7 Derive the score values of each alternative \(A_{i}\).
Utilizing Definition 14, the score values for each alternative can be calculated.
-
Step 8 Gain the final ranking order and select the optimal alternative(s).
By comparing the values obtained in Step 8, the final ranking results can be obtained, and the optimum option can be selected.
6 An numerical example
This section uses a green supplier selection problem adapted from Wan and Dong [60] to demonstrate the applicability of the proposed method.
Shanghai General Motors Company Limited (SGM) is planning to incorporate environmentally friendly features into the product design stage to protect the environment and achieve sustainable development of the social economy. For this reason, SGM wishes to select the most appropriate green supplier for one of the key elements in its manufacturing process. After pre-evaluation, four suppliers remain as candidates for further evaluation. They are Howden Hua Engineering Company (\(A_{1}\)), Sino Trunk (\(A_{2}\)), Taikai Electric Group Company Limited (\(A_{3}\)), and Shantui Construction Machinery Company Limited (\(A_{4}\)). Utilizing principal component analysis, the experts choose the following three independent criteria as evaluation principles: product quality (\(C_{1}\)), pollution control (\(C_{2}\)), and environment management (\(C_{3}\)). According to historical data, the weight vector of the three criteria is \(\varpi = \left( {0.4,0.35,0.25} \right)^{T}\). The results of pairwise comparisons among these three alternatives with respect to the three criteria \(C_{j} \left( {j = 1,2,3} \right)\) are listed as follows in the form of SVTNNs:
6.1 Evaluation steps for green supplier selection problem
The evaluation steps of the proposed method proceed as follows.
-
Step 1 Normalize the SVTNPRs.
Because all the criteria are of the benefit type, the decision information does not need to be normalized. In other words, \(\bar{R}_{j} = \left( {\bar{a}_{ik}^{j} } \right)_{m \times m} = R_{j} = \left( {\tilde{a}_{ik}^{j} } \right)_{m \times m} \left( {k = 1,2, \ldots ,m} \right)\).
-
Step 2 Check the complete consistency or acceptable consistency of SVTNPR \(\widetilde{R}\).
Based on Definition 19, the results obtained by calculating the elements in SVTNPR \(R_{1}\) are as follows: \(a_{31}^{11} + a_{31}^{14} + a_{12}^{11} + a_{12}^{14} + a_{23}^{11} + a_{23}^{14} = 0.4 + 0.6 + 0.3 + 0.7 + 0.4 + 0.7 = 3.1 \ne 3,\) and \(a_{31}^{12} + a_{31}^{13} + a_{12}^{12} + a_{12}^{13} + a_{23}^{12} + a_{23}^{13} = 0.5 + 0.5 + 0.5 + 0.5 + 0.5 + 0.6 = 3.1 \ne 3\); therefore, \(R_{1}\) is not completely consistent. However, \(R_{2}\) and \(R_{3}\) are both completely consistent according to Definition 19. Furthermore, utilizing the score function defined in Definition 14 in Sect. 3.2, we can obtain \(E\left( {\bar{a}_{13}^{1} } \right) = 0.175\), \(E\left( {\bar{a}_{32}^{1} } \right) = 0.100\), \(E\left( {\bar{a}_{12}^{1} } \right) = 0.342\), and \(E\left( {\bar{a}_{11}^{1} } \right) = 0.333\), such that \(\bar{a}_{13}^{1} \prec \bar{a}_{11}^{1}\), \(\bar{a}_{32}^{1} \prec \bar{a}_{11}^{1}\), and \(\bar{a}_{12}^{1} \prec \bar{a}_{11}^{1}\); therefore, according to Definition 20, \(R_{1}\) is not acceptably consistent either.
As a result, the matrix \(R_{1}\) must be modified based on Definitions 19 and 20 until it is completely consistent.
-
Step 3 Improve the consistency degree of SVTNPR \(\bar{R}\).
Assume that the experts are very sure about the first row of SVTNPR \(\bar{R}\); then, based on Definition 18, we can see that the experts are also very sure about the first column of SVTNPR \(\bar{R}\).
Let the elements in the second line be divided by the elements in the first line. The results are not a fixed value, but the three membership degrees satisfy the conditions in Definition 19; therefore, \(\tilde{a}_{23}^{1}\) should be modified as \(\tilde{a}_{23}^{1} { = }\left\langle {\left[ {0.3,0.5,0.5,0.7} \right],\left( {0.7,0,0.3} \right)} \right\rangle\).
-
Step 4 Obtain the completely consistent SVTNPR.
Check every element in the SVTNPR \(R_{1} = \left( {\alpha_{ik}^{1} } \right)_{m \times m}\) and modify all the elements; then, the modified results of \(\overline{R}_{1}\) are as follows:
-
Step 5 Obtain the overall preference information.
Utilizing Eqs. (13) and (16), the overall preference value can be aggregated. It is denoted as matrices \(T_{\text{SVTNWAA}}\) and \(T^{\prime}_{\text{SVTNWGA}}\) which are given as follows:
and
-
Step 6 Calculate the ordering vector of each alternative.
Aggregate each row of SVTNPR \(T_{\text{SVTNWAA}} = \left( {t_{ik} } \right)_{m \times m}\) and \(T^{\prime}_{\text{SVTNWGA}} = \left( {t^{\prime}_{ik} } \right)_{m \times m}\), yielding the matrices \(U = \left( {u_{i} } \right)_{m \times 1}\) and \(U^{\prime} = \left( {u^{\prime}_{i} } \right)_{m \times 1}\), respectively, which are composed of the ordering vector of each alternative.
and
Then, the normalized matrix \(V = \left( {\tilde{u}_{i} } \right)_{m \times 1}^{\text{T}} = \left( {\tilde{u}_{1} ,\tilde{u}_{2} , \ldots ,\tilde{u}_{m} } \right)\) from \(U\) is as follows:
and the normalized matrix \(V^{\prime} = \left( {u^{\prime\prime}_{i} } \right)_{m \times 1}^{\text{T}} = \left( {u^{\prime\prime}_{1} ,u^{\prime\prime}_{2} , \ldots ,u^{\prime\prime}_{m} } \right)\) from \(U^{\prime}\) is as follows:
-
Step 7 Derive the score values of each alternative \(A_{i}\).
The score values of each alternative can be obtained based on Definition 14.
When using the alternative information in \(U\):
When using the alternative information in \(U^{\prime}\):
-
Step 8 Gain the final ranking order and select the optimal alternative(s).
By comparing the values obtained in Step 8, including \(E(A_{1} ) > E(A_{2} ) > E(A_{3} )\) obtained by \(U\) and \(E(A_{2} ) > E(A_{1} ) > E(A_{3} )\) obtained by \(U^{\prime}\), we can get the final ranking orders as \(A_{1} \succ A_{2} \succ A_{3}\) and \(A_{2} \succ A_{1} \succ A_{3}\), respectively. Because of the distinct inherent characteristic of these two operators, it is reasonable that the ultimate ranking results are different.
6.2 Comparative study and discussion
In order to illustrate the rationality and effectiveness of the proposed method, this subsection conducts a comparative study and discussion.
In solving the above example, the preference information of each alternative \(A_{i}\) is integrated directly using the SVTNWAA and SVTNWGA operators, and consistency is not considered. The aggregated matrices can be calculated, respectively, as follows:
and
Then, the corresponding normalized matrices can be calculated, respectively, as follows:
and
Therefore, we can obtain the score values based on Definition 14, and they are \(\dot{E}(A_{1} ) = 0.4996\), \(\dot{E}(A_{2} ) = 0.4645\), and \(\dot{E}(A_{3} ) = 0.4023\) in the normalized matrix \(\dot{V}\), such that the ranking order is \(A_{1} \succ A_{2} \succ A_{3}\); meanwhile, they are \(\dot{E}(A_{1} ) = 0.3015\), \(\dot{E}(A_{2} ) = 0.3145\), and \(\dot{E}(A_{3} ) = 0.2801\) in the normalized matrix \(\dot{V}^{\prime}\), such that the ranking order is \(A_{2} \succ A_{1} \succ A_{3}\). Consequently, we can see that a different result is produced when the consistency of preference relations is not considered. This also indicates that our approach is reasonable.
7 Conclusion
This paper developed a novel single-valued trapezoidal neutrosophic MCDM method based on SVTNPRs. First of all, in order to overcome the disadvantages of existing operations and comparison methods for SVTNNs, which are not always in accordance with real MCDM situations, we defined some new operations and a new comparison method and explored their properties; second, we developed the SVTNWAA operator and SVTNWGA operator to aggregate decision information. And then, we constructed the SVTNPRs and defined both the complete consistency and acceptable consistency. Finally, we outlined a method for MCDM problem solving with SVTNPRs and applied a numerical example to verify the effectiveness of our method. Future research should take into account unknown weight information, which exists widely in real life; additionally, DMs’ preferences on alternatives might be missing, and the problem of tackling incomplete preference information also represents an important issue for further study. Also, it is worthwhile to investigate SVNPR when the neutrosophic components are partially dependent and partially independent. Finally, it is worthwhile to propose new comparison methods and integrated methodologies to deal with MCDM problems with SVTNNs.
References
Tian Z-P, Wang J, Wang J-Q, Zhang H-Y (2016) Simplified neutrosophic linguistic multi-criteria group decision-making approach to green product development. Group Decis Negot. doi:10.1007/s10726-016-9479-5
Humphreys P, McIvor R, Chan F (2003) Using case-based reasoning to evaluate supplier environmental management performance. Expert Syst Appl 25(2):141–153
Govindan K, Rajendran S, Sarkis J, Murugesan P (2015) Multi criteria decision making approaches for green supplier evaluation and selection: a literature review. J Clean Prod 98:66–83
Keshavarz Ghorabaee M, Zavadskas EK, Amiri M, Esmaeili A (2016) Multi-criteria evaluation of green suppliers using an extended WASPAS method with interval type-2 fuzzy sets. J Clean Prod 137:213–229
Chai J, Liu JNK, Ngai EWT (2013) Application of decision-making techniques in supplier selection: a systematic review of literature. Expert Syst Appl 40(10):3872–3885
Noci G (1997) Designing ‘green’ vendor rating systems for the assessment of a supplier’s environmental performance. Eur J Purch Supply Manag 3(2):103–114
Handfield R, Walton SV, Sroufe R, Melnyk SA (2002) Applying environmental criteria to supplier assessment a study in the application of the analytical hierarchy process. Eur J Oper Res 141(1):70–87
Büyüközkan G, Çifçi G (2011) A novel fuzzy multi-criteria decision framework for sustainable supplier selection with incomplete information. Comput Ind 62(2):164–174
Büyüközkan G, Çifçi G (2011) Evaluation of the green supply chain management practices: a fuzzy ANP approach. Prod Plan Control 23(6):1–14
Hashemi SH, Karimi A, Tavana M (2015) An integrated green supplier selection approach with analytic network process and improved Grey relational analysis. Int J Prod Econ 159:178–191
Yeh W-C, Chuang M-C (2011) Using multi-objective genetic algorithm for partner selection in green supply chain problems. Expert Syst Appl 38(4):4244–4253
Wu J (2015) An intuitionistic fuzzy judgement matrix and TOPSIS integrated multi-criteria decision making method for green supplier selection. J Intell Fuzzy Syst 28(1):117–126
Vachon S, Klassen RD (2006) Green project partnership in the supply chain: the case of the package printing industry. J Clean Prod 14(6–7):661–671
Qin J-D, Liu X-W, Pedrycz W (2017) An extended TODIM multi-criteria group decision making method for green supplier selection in interval type-2 fuzzy environment. Eur J Oper Res 258(2):626–638
Awasthi A, Chauhan SS, Goyal SK (2010) A fuzzy multicriteria approach for evaluating environmental performance of suppliers. Int J Prod Econ 126(2):370–378
Sen DK, Datta S, Mahapatra SS (2016) A TODIM-based decision support framework for g-resilient supplier selection in fuzzy environment. Asia Pac J Oper Res 33(05):1650033
Zadeh LA (1965) Fuzzy sets. Inf Control 8(3):338–353
Chan H-K, Wang X-J, Raffoni A (2014) An integrated approach for green design: life-cycle, fuzzy AHP and environmental management accounting. Br Account Rev 46(4):344–360
Kumar D, Rahman Z, Chan FTS (2016) A fuzzy AHP and fuzzy multi-objective linear programming model for order allocation in a sustainable supply chain: a case study. Int J Comput Integr Manuf. doi:10.1080/0951192x.2016.1145813
Büyüközkan G, Güleryüz S (2016) A new integrated intuitionistic fuzzy group decision making approach for product development partner selection. Comput Ind Eng 102:383–395
Atanassov KT (1986) Intuitionistic fuzzy sets. Fuzzy Sets Syst 20(1):87–96
Mendel JM, John RI, Liu F (2007) Interval type-2 fuzzy logic systems made simple. IEEE Trans Fuzzy Syst 14(6):808–821
Yu S-M, Wang J, Wang J-Q (2017) An interval type-2 fuzzy likelihood-based MABAC approach and its application in selecting hotels on a tourism website. Int J Fuzzy Syst 19(1):47–61
Ghorabaee MK, Amiri M, Sadaghiani JS, Zavadskas EK (2015) Multi-criteria project selection using an extended VIKOR method with interval type-2 fuzzy sets. Int J Inf Technol Decis Mak 14(05):993–1016
Ghorabaee MK, Amiri M, Sadaghiani JS, Goodarzi GH (2014) Multiple criteria group decision-making for supplier selection based on COPRAS method with interval type-2 fuzzy sets. Int J Adv Manuf Technol 75(5–8):1115–1130
Peng H-G, Wang J-Q (2016) Hesitant uncertain linguistic Z-numbers and their application in multi-criteria group decision-making problems. Int J Fuzzy Syst. doi:10.1007/s40815-016-0257-y
Yu S-M, Wang J, Wang J-Q (2016) An extended TODIM approach with intuitionistic linguistic numbers. Int Trans Oper Res. doi:10.1111/itor.12363
Peng H-G, Wang J-Q, Chen P-F (2016) A linguistic intuitionistic multi-criteria decision-making method based on the Frank Heronian mean operator and its application in evaluating coal mine safety. Int J Mach Learn Cybern. doi:10.1007/s13042-016-0630-z
Smarandache F (1999) A unifying field in logics: neutrosophic logic. Neutrosophy, neutrosophic set, neutrosophic probability. American Research Press, Rehoboth, pp 1–141
Smarandache F (1998) Neutrosophy: neutrosophic probability, set, and logic. American Research Press, Rehoboth, pp 1–105
Smarandache F (2008) Neutrosophic set—a generalization of the intuitionistic fuzzy set. Int J Pure Appl Math 24(3):38–42
Smarandache F (2016) Degree of dependence and independence of the (sub)components of fuzzy set and neutrosophic set. Neutrosophic Sets Syst 11:95–97
Ji P, Wang J-Q, Zhang H-Y (2016) Frank prioritized Bonferroni mean operator with single-valued neutrosophic sets and its application in selecting third-party logistics providers. Neural Comput Appl. doi:10.1007/s00521-016-2660-6
Zhang H-Y, Ji P, Wang J, Chen X-H (2017) A novel decision support model for satisfactory restaurants utilizing social information: a case study of TripAdvisor.com. Tour Manag 59:281–297
Ji P, Zhang H-Y, Wang J-Q (2016) A projection-based TODIM method under multi-valued neutrosophic environments and its application in personnel selection. Neural Comput Appl. doi:10.1007/s00521-016-2436-z
Peng H-G, Zhang H-Y, Wang J-Q (2016) Probability multi-valued neutrosophic sets and its application in multi-criteria group decision-making problems. Neural Comput Appl. doi:10.1007/s00521-016-2702-0
Peng J-J, Wang J-Q, Wu X-H (2016) An extension of the ELECTRE approach with multi-valued neutrosophic information. Neural Comput Appl. doi:10.1007/s00521-016-2411-8
Wang J-Q, Yang Y, Li L (2016) Multi-criteria decision-making method based on single-valued neutrosophic linguistic Maclaurin symmetric mean operators. Neural Comput Appl. doi:10.1007/s00521-016-2747-0
Ye J (2015) Some weighted aggregation operators of trapezoidal neutrosophic numbers and their multiple attribute decision making method. http://fs.gallup.unm.edu/SomeWeightedAggregationOperators.pdf
Deli I, Şubaş Y (2015) Some weighted geometric operators with SVTrN-numbers and their application to multi-criteria decision making problems. J Intell Fuzzy Syst. doi:10.3233/jifs-151677
Deli I, Şubaş Y (2016) A ranking method of single valued neutrosophic numbers and its applications to multi-attribute decision making problems. Int J Mach Learn Cybern. doi:10.1007/s13042-016-0505-3
Biswas P, Pramanik S, Giri BC (2016) Value and ambiguity index based ranking method of single-valued trapezoidal neutrosophic numbers and its application to multi-attribute decision making. Neutrosophic Sets Syst 12:127–138
Thamaraiselvi A, Santhi R (2016) A new approach for optimization of real life transportation problem in neutrosophic environment. Math Probl Eng. doi:10.1155/2016/5950747
Liang R-X, Wang J-Q, Li L (2016) Multi-criteria group decision making method based on interdependent inputs of single valued trapezoidal neutrosophic information. Neural Comput Appl. doi:10.1007/s00521-016-2672-2
Xu G-L, Wan S-P, Wang F, Dong J-Y, Zeng Y-F (2015) Mathematical programming methods for consistency and consensus in group decision making with intuitionistic fuzzy preference relations. Knowl Based Syst 98:30–43
Zhang Y-B, Li K-W, Wang Z-J (2017) Prioritization and aggregation of intuitionistic preference relations: a multiplicative-transitivity-based transformation from intuitionistic judgment data to priority weights. Group Decis Negot 26(2):409–436
Meng F, Chen X (2015) An approach to incomplete multiplicative preference relations and its application in group decision making. Inf Sci 309(C):119–137
Jin F, Ni Z, Chen H, Li Y (2016) Approaches to group decision making with intuitionistic fuzzy preference relations based on multiplicative consistency. Knowl Based Syst 97(C):48–59
Xu Z, Liao H (2015) A survey of approaches to decision making with intuitionistic fuzzy preference relations. Knowl-Based Syst 80(C):131–142
Orlovsky S (1978) Decision-making with a fuzzy preference relation. Fuzzy Sets Syst 1(3):155–167
Saaty TL (1986) Axiomatic foundation of the analytic hierarchy process. Manag Sci 32(7):841–855
Xu Z (2007) Intuitionistic preference relations and their application in group decision making. Inf Sci 177(11):2363–2379
Xu Z, Liao H (2015) A survey of approaches to decision making with intuitionistic fuzzy preference relations. Knowl Based Syst 80:131–142
Wang J-H, Hao J-Y (2006) A new version of 2-tuple fuzzy linguistic representation model for computing with words. IEEE Trans Fuzzy Syst 14(3):435–445
Dubois D, Prade H (1983) Ranking fuzzy numbers in the setting of possibility theory. Inf Sci 30(3):183–224
Wang H-B, Smarandache F, Zhang Y, Sunderraman R (2010) Single valued neutrosophic sets. Multispace Multistruct 4:410–413
Deli I, Şubaş Y (2014) Single valued neutrosophic numbers and their applications to multicriteria decision making problem. viXra preprint viXra 14120012
Wang Y-M (2009) Centroid defuzzification and the maximizing set and minimizing set ranking based on alpha level sets. Comput Ind Eng 57(1):228–236
Şahin R (2014) Multi-criteria neutrosophic decision making method based on score and accuracy functions under neutrosophic environment. arXiv preprint arXiv:1412.5202
Wan S-P, Dong J-Y (2015) Power geometric operators of trapezoidal intuitionistic fuzzy numbers and application to multi-attribute group decision making. Appl Soft Comput 29:153–168
Acknowledgements
This work was supported by the National Natural Science Foundation of China (No. 71571193). The authors also would like to express appreciation to the anonymous reviewers and editors for their helpful comments that improved the paper.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that there is no conflict of interest regarding the publication of this paper.
Rights and permissions
About this article
Cite this article
Liang, Rx., Wang, Jq. & Zhang, Hy. A multi-criteria decision-making method based on single-valued trapezoidal neutrosophic preference relations with complete weight information. Neural Comput & Applic 30, 3383–3398 (2018). https://doi.org/10.1007/s00521-017-2925-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-017-2925-8