In this section, basic definitions of neutrosophic sets, single-valued neutrosophic sets and rough neutrosophic sets are recalled.
2.2. Rough Neutrosophic Sets
Two basic components in rough set theory, the crisp set and equivalence relation, are the mathematical basis of rough sets. The fundamental idea of a rough set is an approximation of the set known as the lower and upper approximation. Rough neutrosophic sets [
5] are generalizations of rough fuzzy sets [
53] and rough intuitionistic fuzzy sets [
54].
Definition 3 ([
52])
. Let be a non-null set and be an equivalence relation on . Let be a neutrosophic set in with the membership function , indeterminacy function and non-membership function . The lower and the upper approximations of in the approximation denoted by and are defined respectively as:
where
Therefore:
Operators
and
present the “min” and the “max’’, respectively.
,
and
are membership, indeterminacy and non-membership of
with respect to
. It is obvious that
and
are two neutrosophic sets in
. Therefore, the neutrosophic set mapping
,
:
presents the lower and upper rough neutrosophic sets approximation operators, and the pair
is the rough neutrosophic set in
. According to Definition 3, it is obvious that
and
have constant membership on the equivalence clases of
if
,
is a definable neutrosophic set in the approximation
, and according to [
55], it is easy to prove that the zero neutrosophic set
and unit neutrosophic sets
are definable neutrosophic sets.
If
is a rough neutrosophic set in
, the rough neutrosophic of
is a rough neutrosophic set denoted by
, where
represent the complements of neutrosophic sets of
, respectively.
If
and
are two rough neutrosophic sets in
, then according to [
5], the definitions are given as:
If are rough neutrosophic sets in , then the following propositions are as follows:
Proposition 1. - i.
- ii.
- iii.
- iv.
- v.
- vi.
Proposition 2.
For any two neutrosophic sets, De Morgan’s laws are satisfied:
- i.
- ii.
Proposition 3.
Ifandare two neutrosophic sets insuch that, then:
- i.
- ii.
Proposition 4.
For any rough neutrosophic set:
- i.
- ii.
- iii.
2.3. Rough Neutrosophic Symmetric Cross Entropy Measure
In this section, a new rough neutrosophic symmetric cross entropy measure is defined to measure the difference between two rough neutrosophic variables.
Definition 4 ([
56])
. Let be a probability distribution on , where , and let be the other probability distribution on , where . The cross entropy between and probability distributions is described as the Kullback–Leibler distance [57], and is determined as follows:
which is not a distance in the formal meaning, since it does not satisfy the symmetry, i.e.,
and the triangle inequality. Although, the symmetry can be achieved by adding
to
, as:
when
, it can be assumed that
and
, and Equation (10) can be defined as:
Definition 5. Letandbe two neutrosophic sets in the universe of discourse,,. According to Equation
(10), the neutrosophic symmetric cross entropy of from can be expressed as follows: The neutrosophic symmetric cross-entropy of and satisfies the following properties:
;
;
.
Definition 6. Letandbe any two rough neutrosophic sets in. Then, the rough neutrosophic symmetric cross-entropy ofandis denoted byand defined as follows: If
and
, and vice versa,
cannot be defined, and taking into the account Equation (13), the authors propose a modified rough neutrosophic symmetric cross entropy as follows:
If
then (15) can be written as:
Wang et al. [
58] defined operational laws for two single-valued neutrosophic numbers
and
as follows:
, .
According to [
59], the sum of membership, indeterminacy and non-membership with respect to
of the lower and upper rough neutrosophic sets are defined as:
Definition 7. Considering Wang et al.’s operational laws and Mondal et al.’s definition of relationship of lower and upper rough neutrosophic sets, the operational laws for two rough neutrosophic setsandcan be written as follows: In the following, according to Equation (16), and Definition 7, the authors developed the rough neutrosophic symmetric cross entropy:
Example 1. Let A and B be two rough neutrosophic sets in Z, given by and. Using (21), the rough neutrosophic symmetric cross entropy of value A and B is obtained as.
Theorem 1. Rough neutrosophic symmetric cross entropyfor any two rough neutrosophic sets A, B satisfies the following properties:
iff , ,
Proof. (i) For all values of
,
Similarly,
Therefore, and it completes the proof.
Therefore,
. Hence, the proof is completed.
(iii) As is,
and the same can be applied to the rest of Expression (21). Therefore, , and the proof is completed. □
2.5. VIKOR Method
The VIKOR method is a multicriteria method developed to solve decision-making problems that focus on the ranking and selection of a set of alternatives evaluated by utilizing multiple criteria. The method is based on the measure of closeness to the ideal solution (distance-to-target) [
25]. Based on the abovementioned factors, the methodology of VIKOR consists of the following steps [
60]:
Step 1: Definition of the alternatives , ,, . is the weight of the th criterion, denoting the relative importance of the criteria, where ; is the number of alternatives, and is the number of criteria. The rating of the th criterion is denoted by for alternative .
Step 2: Define the best
and the worst
values of criterion functions,
;
Step 3: Calculation of the values
and
(utility measure and regret measure), that emphasizes the maximum group utility and selecting minimum among the maximum individual regrets, respectively;
, using two relations:
Step 4: Calculation of the values
,
, using the relation:
where,
,
,
,
, and
,
is a weight of maximum group utility, and
is the weight of the individual regret.
Step 5: Ranking the alternatives by values and in decreasing order.
2.7. Proposed Modified Rough Neutrosophic VIKOR Based on RNS Cross Entropy
In this section, a VIKOR strategy under a rough neutrosophic environment based on RNS cross entropy is proposed.
Let be a set of alternatives, and be a set of criteria. Assume that is the weight vector of the criteria, where . Let be a set of experts. The newly developed methodology consists of the following steps:
Step 1.Definition of comparison matrix and aggregation of criteria weights.
Experts define comparison matrices of criteria using rough neutrosophic numbers and linguistic values. Criteria weights are then calculated using (25) and (26), which provide their crisp values. Furthermore, the aggregated comparison matrix is calculated using the geometric mean operator [
61]:
Step 2.Determination of decision matrix.
Let
be the
th rough neutrosophic decision matrix, where assessment of the alternative
is obtained by expert
with respect to criteria
. The
pth decision matrix marked by
is defined as:
where
Step 3.Aggregation of decision matrices.
Let
be a collection of rough neutrosophic numbers and
be the weight structure of rough neutrosophic numbers
. Then, the weighted rough neutrosophic geometric mean (WRNGMO) is defined as [
62]:
Let A and B be two rough neutrosophic sets in relation to
based on
. Hence,
is multiplied as [
63]:
Taking into account (31) and (32), and considering
as rough neutrosophic sets, the expression of the weighted rough neutrosophic geometric mean operator (WRNGMO) can be proposed as:
Exemple 3. Let two rough neutrosophic numbers be,andandthen, according to expression (33): Step 4.Definition of benefit type criteria and cost type criteria.
Benefit type of criteria:
Step 5.Definition of RN utility measure and RN regret measure.
In decision making processes, there are cost type and benefit type criteria. Therefore, to avoid different physical dimensional units, the considered criteria weight values need to be normalized.
Let
and
be two rough neutrosophic numbers in
where
, and the normalization can be calculated as follows [
58]:
is the benefit type of criteria, and is the cost type of criteria for each alternative.
Using the operational law defined in Defintion 7, the normalization can be calculated as follows:
where
and
and
represent the utility and regret measure of the alternative
, respectively, under the rough neutrosophic environmnet. The expressons for
and
are calcualted by mulitplying weights and normalizations, derived in the previous step, as follows:
Step 6.Calculating the RNS cross entropy.
Determination of the RNS cross entropy starts with the ideal alternative which is defined from aggregated decision matrix , compatible with the benefit criteria, and , which corresponds to the cost criteria, where
Considering proposed weighted RNS cross entropy (22) and criteria weights defined in (27) and (28), weighted RNS cross entropy is calculated between all alternatives and the ideal alternative, as presented below:
The smaller the value is, the closer the alternative is to the ideal solution. Hence, the priority ranking is obtained according to an increasing sequence of the values of (
Step 7.Calculation of VIKOR index.
Hereby, the VIKOR index is defined, taking into account rough neutrosophic characteristics of utility and regret measure, and the decision-making mechanism coefficient
is considered from 0 to 1.
where
,
,
,
, and
is the decision-making mechanism coefficient [
64]. For the maximum group utility,
and for the minimum group utility,
. The median is taken as
[
50].
The compromise alternative is
, which has the minimum
value when two conditions are satisfied [
65]:
Condition 1. , whereis the second ranked alternative byin the ranking list, and.
Condition 2. Ifis also ranked first byand, then it is the most satisfying in the decision-making process.
In the case when one of the conditions is not satisfied, the compromise alternative is obtained by following conditions:
Condition 3. andare considered compromised if Condition 2 is not satisfied, or
Condition 4. are considered compromised if Condition 1 is not satisfied, andis defined by.
Step 8.Priority ranking.
Priority ranking is defined by
,
and
. Smaller values indicate the higher priority of the alternative. Furthermore, in
Figure 1, the proposed methodology is illustratively presented in detail, throughout the framework.