Covering soft rough sets and its topological properties with application

In this article, we introduce the notion of the complementary soft neighborhood and present three kinds of covering soft rough set (CSR\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {CSR}$$\end{document}) models. The basic properties of these models are investigated. The relationships among these models are also discussed. Moreover, we establish the topological approach to CSR\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {CSR}$$\end{document} say, Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-topological spaces (i.e., Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-TS). Hence, the topological properties for Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-TS models such as Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-open sets, Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-closed sets, Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-interior, Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-closure, Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-boundary, Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-neighborhood and Δ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta $$\end{document}-limit point are studied and the relationships between them are given. Finally, we make use of an algorithm for these proposed models to deal with uncertainties for solving the MGDM problems using the constructed topologies.


Introduction
The concept of rough set theory (briefly, RST) was launched by Pawlak in 1982(Pawlak 1982, 1985 as a tool for representing and computing information in data tables. The fuzzy rough set theory (briefly, FRS) was introduced by Dubois and Prade (1990) as a combination between fuzzy set and rough set which offers a wide variety of techniques to analyze imprecise data and the RST. Since the uncertainty and incompleteness of knowledge are widespread phenomena in information systems, the FRS is a helpful approach to handle such situations. The first step in studying rough set theory was based on equivalence relations, but various extensions were developed to different mathematical concepts such as a binary relation, or a neighborhood system of a topological space and a covering (Atef et al. 2020;Alcantud and Zhan 2020;El Atik et al. 2021;Herawan et al. 2010;Hu et al. 2010;Huang et al. 2011;Jensen and Shen 2004;Liu and Zhu 2008;Pal and Mitra 2004;Qian et al. 2009;Yang and Li 2006;Yao 2010;Zhang et al. 2004;Sun et al. 2017;Wu and Zhang 2004;Yeung et al. 2015;Ziarko 1993 Menoufia, Egypt 1987, 1988Yao 1998;Yao and Yao 2012;Couso and Dubois 2011;Bonikowski et al. 1998;Zhu 2007;Zhu and Wang 2003, 2007, 2012Tsang et al. 2008;Xu and Zhang 2007;Liu and Sai 2009).
In 2007, Deng et al. (2007) developed a new model for fuzzy rough set based on the concepts of both fuzzy covering and binary fuzzy conjunction and fuzzy implication in the framework of lattice theory. In 2008, Li et al. (2008) focused on studying the covering based rough set models via the fuzzy covering that are constructed by the lower and upper fuzzy rough approximation operators. In 2012, Ma (2012) defined the notion of complementary neighborhood of an element in the universal set and according to this notion and the classical neighborhood notion; she showed the equivalency of both notions to some types of covering rough sets.
Later, in 2016, Ma (2016) introduced the concepts of fuzzy β-covering and fuzzy β-neighborhood and then constructed two kinds of fuzzy covering-based rough set models. It should be noted that the concept of fuzzy covering is a special case of fuzzy β-covering. Yang and Hu (2017), Yang and Hu (2019) introduced three types of fuzzy covering based rough set models which are generalizations of Ma's models, and also they proposed four types of fuzzy neighborhood operators based on a fuzzy β-covering using the notions fuzzy β-minimal and β-maximal descriptions. The concept of fuzzy neighborhood system of an object based on a given fuzzy covering was defined and discussed by D'eer et al. (2017).
We were in much need to the soft set theory (briefly, SST) notion considered by Molodtsov (1999) since many mathematical models cannot use the previous classical tools such as theory of fuzzy sets , theory of intuitionistic fuzzy sets and theory of rough sets because of the difficulties of some types of uncertainties. The absence of all restrictions on the approximate description in soft set theory makes this theory easily applicable in practice in different topics. A combination of fuzzy sets with soft sets made by Maji et al. (2002) initiated an important model. Several authors have applied this theory in their research (Alcantud 2020;Ali 2011;Feng et al. 2010;Li and Xie 2014;Min 2020;Zhan and Wang 2019).
In 2010, Feng et al. (2011) introduced soft approximation spaces, soft rough approximations and soft rough sets based on granulation structures. Later, Feng (2011) gave an application of soft rough approximation in multi-criteria group decision-making problems. Yüksel et al. (2014), Yüksel et al. (2015) established a soft covering approximation space from a covering soft set and then defined the soft minimal description of an object, and explored an application of soft covering based rough set. Li et al. (2015) investigated parameter reductions of a soft covering. Riaz et al. (2019) brought out topological structures of soft rough sets. Another extension was made by . They introduced a soft rough covering (CSR) by means of soft neighborhoods and then used it to improve decision making in a multi-criteria group environment. A new approach called modified soft rough set, for studying roughness through soft sets in order to find approximations of a set was done by Shabir et al. (2013).
The objective of this article is to introduce new types of soft neighborhoods and use them to construct novel models of CSR which are of better quality than the existence models (i.e., growing the lower approximation and lowering the upper approximation of Feng's, Riaz's and Zhan's models). The relationships between them being clarified. In addition, we discuss some topological properties of the four topologies introduced based on the CSR approach. Finally, we explain the relationships among these kinds and establish a numerical example to solve MGDM problems. The body of this paper is implemented here. Section 2 gives basic terminologies. Section 3 describes three types of CSR by using the notions of soft neighborhoods and complementary soft neighborhoods. In Sect. 4, we establish the topological approach for CSR models and study their properties. In Sect. 5, we solve a real problem through the proposed study. We write the aim of this article in Sect. 6.

Preliminaries
a quick survey of the notions which will be used in the coming sections.
Definition 2.1 Yang and Hu (2019) Let ϒ be a universe of discourse, and E be a finite set of relevant parameters regarding ϒ. The pair S = (F, A) is a soft set over ϒ, when A ⊆ E and F : A → P(ϒ) (i.e., F is a set-valued mapping from the subset of attributes A to ϒ and P(ϒ) denotes the set of all subsets of ϒ).
If F (L) = F ⊕ (L), then L is called Feng-approximation soft rough sets(F − ASR for short), otherwise it is definable.
In Shabir et al. (2013), Shabir et al. introduced the concept of modified soft rough sets (MSR-sets for short) as follows.
If S (L) = S ⊕ (L), then L is called Shabir-approximation modified soft rough sets (S − AMSR for short), otherwise it is definable.
Definition 2.5 Yüksel et al. (2014Yüksel et al. ( , 2015 Let (ϒ, F, A) be an SCAS. For each g ∈ ϒ, define a soft minimal description of g as follows.
Definition 2.7 Li et al. (2015) Let (ϒ, F, A) be an SCAS. For each L ⊆ ϒ, define the Li-lower and Li-upper approximations, respectively, as follows.
If L (L) = L ⊕ (L), then L is called Li-covering soft rough sets (L − CSR for short), otherwise it is definable. (2019) Let (ϒ, F, A) be an SCAS. For each g ∈ ϒ, define a soft neighborhood of g as follows.
, then L is called Zhan-covering soft rough sets (CSR 1 for short), otherwise it is definable.

Covering soft rough sets
The aim of this section is to extent the idea of Zhan's method given in [61]. This will be done via the complementary soft neighborhood. Also, we generalize it using two known neighborhoods through combining between Zhan's and complementary neighborhoods. Thus three new kinds of CSR are investigated. The relevant characteristics are also discussed.
define three kinds of the soft neighborhood of g as follows. Table 1.

Example 3.2 Let (ϒ, F, A) be an SCAS and (F, A) be a soft set given as listed in
Compute the four types of soft neighborhoods as the following.
In the following, we can obtain the accuracy measure for the four types of covering soft rough sets in Definitions 2.9 and 3.3 as follows.
The following example shows the differences between these types of measures.
Example 3.6 Consider again Example 3.4. Calculate the accuracy measures as follows.
From the above example, it is easy to see that CSR 3 is more accurate than the others. The following two Theorems 3.7 and 3.8 for CSR 2 , CSR 3 , and CSR 4 hold. We prove them in the case of CSR 2 only and the others are similar.
Theorem 3.7 Let (ϒ, F, A) be an SCAS. If L 1 , L 2 ⊆ ϒ, then we have the following relations hold.
Proof The proof is similar to Theorem 3.7.
Definition 3.9 Let (ϒ, F, A) be an SCAS. For each L ⊆ ϒ, define the positive, negative and boundary region for the three kinds CSR of L as follows.
The next remark explains the comparisons between the given approximations by using the boundary region.

Remark 3.11
The goal of approximations is to reduce the boundary zone as possible to get a reliable solution. Here, by the above Example 3.10, B 3 (L) is least than the others (i.e., B 1 (L), B 2 (L) and B 4 (L)) . This means that CSR 3 is more accurate than the others.
Next, we present new kind of a degree of soft rough membership as follows. (ϒ, F, A) be an SCAS. For each L ⊆ ϒ, define three new kinds of a degree of soft rough membership of L as follows.

Definition 3.12 Let
Example 3.13 Consider Example 3.2. If L = {g 1 , g 3 , g 5 }, then we get the following values.
The next two Theorems 3.14 and 3.15 are true for D 2 (h, L), D 3 (h, L), and D 4 (h, L). We prove them for D 2 (h, L) only and for the other cases the proofs are similar.
Theorem 3.14 Let (ϒ, F, A) be an SCAS. If L ⊆ ϒ and h ∈ ϒ, then we have the following characteristics.
(3) It is clear by Definition 3.12. (ϒ, F, A) be an SCAS. If L ⊆ ϒ and h ∈ ϒ, then we have the following items.
Next, let us show the relationship between the different CSR models. These are clarified by the following propositions. (ϒ, F, A) be an SCAS. If L ⊆ ϒ and h ∈ ϒ, then we have the following characteristics.

Topological approach to CSR
This section deals with all the definitions of four types of topologies come from CSR models called Δ − T S and explored their relevant properties.
Definition 4.1 Let (ϒ, F, A) be an SCAS of ϒ. For each g ∈ ϒ, define four kinds of topologies as follow. (1) It is obvious that ϒ and ∅ are in T 1 .
(2) Assume that {L l : l ∈ I } be in T 1 and let g ∈ ∪L l .

From the above relations, T Δ (L) is called Δ-topology, then (ϒ, T Δ (L), A) is called Δ-topological space (Δ-T S).
Now, let us show in details these topologies when we consider Example 4.2.
Example 4.2 Consider Example 3.2, then the four topologies are listed below. The steps for calculating the first type of topology is as follows.
Proof The proof is similar to the one in Theorem 4.7. ( Proof We shall only prove the theorem taking Δ = {1}. The other cases are almost similar. (1) Suppose that L ⊆ ϒ.
As a consensus of the above theorem we have the following remark. Definition 4.12 Let (ϒ, T Δ (L), A) be a Δ-T S. For each g ∈ ϒ and L ⊆ ϒ. We say L is a Δ-neighborhood of g if ∃L g ∈ T Δ with g ∈ L g ⊆ L.
Definition 4.13 Let (ϒ, T Δ (L), A) be a Δ-T S. For each g ∈ ϒ and L ⊆ ϒ. We say L is a Δ-limit point of g if every Δ-open set T containing g, it is also contains a point of L different from g, i.e., Also, the set of all Δ-limit points of L is said to be the derived set of L and is denoted by L D .
It is an easy target to prove the following theorem, so we omit it. (1) C Δ (L) = L ∪ L D .
Next, we discuss the relationships between these different kinds of topologies as follows.

Proposition 4.15 Let (ϒ, T Δ (L), A) be a Δ-T S.
Then for each L ⊆ ϒ, we have the following relations.

An application by using the proposed method
Let us presume that {g 1 , g 2 , g 3 , ..., g r } be r engineers (alternatives) and {ν 1 , ν 2 , ..., ν m } be m attributes connected to a company which needs to offer a job for an engineering. Thus we use the beauty of soft set to evaluate the optimal engineer through our suggested covering method. So, we establish the MGDM with (ϒ, F, A), where ϒ = {g 1 , g 2 , g 3 , ..., g r } and A = {ν 1 , ν 2 , ..., ν m }. According to the presented models, we built the following algorithm to select the best.
The clarified steps in Algorithm 1 are summarized by the following numerical example.
Example 5.1 Let us consider the engineers form a set ϒ = {g 1 , g 2 , g 3 , ..., g 40 } and the given parameters are presented by the set A = {ν 1 , ν 2 , ..., ν 4 }. These parameters corresponding to the following features, respectively (experience, skills, personality and self-confidence). So, the following steps are obtained.
Step 1 Establish the SC (F, A) as in Table 2.
Step 2 Construct the proficient estimation P = (M, P) and let the output of this estimation be Y . Then M(P i ) = Y i (∀i ∈ {1, 2, 3}), and Table 3 contains the data gained by the experts.
Step 3 Compute the lower and upper approximation for the two soft sets P • = (M • , P) and P Tables 4 and 5.
Also, we obtain the functions for the other two soft set as follows.
Hence, the following results are obtained.
Step 5 Assume that (L (Low gratification), M (Medium gratification), and H (High gratification)) be the parameter set through experts. Thus we can establish Table   Table 6 Table for Hence, the ordering of the alternatives are as follows. Let the select outcome is Δ(g r ) = L (g r ) + M (g r ) + H (g r ). So, we have Therefore, the 3th engineering is the best among others.

Comparative analysis
The prime objective of this work is to growing the lower approximation and lowering the upper approximation of , Feng (2011) and Riaz et al. (2019) as seen in Example 3.4. Now, we intend to clarify the rapprochement between Zhan and Alcantud (2019), Feng (2011), Riaz et al. (2019) and our's, the sorting outcomes of these models are set in Table 7.
In view of Table 7, the optimal alternative is the same (i.e., g 3 ) among all distinct methods that are Feng's method Feng (2011), Zhan's method Zhan and Alcantud (2019) and Riaz's method Riaz et al. (2019), this means that our approach is credible and effective, with vantage to our's in increasing the lower approximation and decreasing the upper approximation which lead to decrease the boundary region as in .

Conclusion
The aim of the employed techniques is to grow the lower approximation and lower the upper approximation depending very strongly on property soft neighborhoods and uses them to construct models of CSR. Also, we studied the relations between these models. Then we turned our attention to study some of the topological properties based on the CSR Zhan's approach Zhan and Alcantud (2019) g 3 ≥ g 1 ≈ g 6 ≥ g 2 ≥ g 4 ≥ g 5 Feng's approach Feng (2011) g 3 > g 5 ≈ g 6 > g 4 > g 1 > g 2 Riaz's approach Riaz et al. (2019) g 3 > g 5 ≈ g 6 > g 4 > g 1 > g 2 Our approach g 3 > g 5 ≈ g 6 ≈ g 1 ≈ g 2 ≈ g 4 approach. Finally, we utilized the algorithm for the proposed model to deal with MGDM problems.
Author Contributions MA, SN and AN contributed to conceptualization, methodology, investigation, writing-original draft preparation, writing-reviewing and editing.
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). Not applicable.
Data Availability Enquiries about data availability should be directed to the authors.

Conflict of interests
The authors declare that they have no conflict of interest.
Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors.
Informed consent Informed consent was obtained from all individual participants included in the study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecomm ons.org/licenses/by/4.0/.