The theory of rough set represents a non-statistical methodology for analyzing ambiguity and imprecise information. It can be characterized by two crisp sets, named the upper and lower approximations that are used to determine the boundary region and accurate measure of any subset. This article endeavors to achieve the best approximation and the highest accuracy degree by using the minimal structure approximation space MSAS via ideal 𝒥. The novel approach (indicated by 𝒥MSAS) modifies the approximation space to diminish the boundary region and enhance the measure of accuracy. The suggested method is more accurate than Pawlak’s and EL-Sharkasy techniques. Via illustrated examples, several remarkable results using these notions are obtained and some of their properties are established. Several sorts of near open (resp. closed) sets based on 𝒥MSAS are studied. Furthermore, the connections between these assorted kinds of near-open sets in 𝒥MSAS are deduced. The advantages and disadvantages of the proposed approach compared to previous ones are examined. An algorithm using MATLAB and a framework for decision-making problems are verified. Finally, the chemical application for the classification of amino acids (AAs) is treated to highlight the significance of applying the suggested approximation.
Idealminimal structure spacesrough set theoryapproximation spacesIntroduction
Topological structures and their generalizations are of crucial importance in data analysis, which have manifested in different fields, for example, in physics [1], chemistry [2], medicine [3], soft set theory [4], etc. Lashin et al. [5] used topological notions to study different issues in rough set theory in order to generalize Pawlak’s concepts [6] in different applications and integrate the concepts of rough and fuzzy sets. Topological structures were applied in rough sets to improve evolutionary-based feature selection technique using the extension of knowledge [7], decision making of COVID-19 [8,9], and enhanced feature selection based on integration containment neighborhoods rough set approximations and binary honey badger optimization [10]. Several articles [11–19] extended the application fields of Pawlak’s model. Popa et al. introduced the minimal structure spaces [20], as a generalization of topological spaces to analyze information systems. Various consequences of minimal spaces can be viewed in [21]. EL-Sharkasy [22,23] studied several sorts of sets in minimal structure spaces with some of their characterizations. The notion of the ideal 𝒥, which is a nonempty class of sets of Λ fulfills the finite additivity and the hereditary property, is famed in the study of topological problems and their other usages defined by Kuratowski [24]. The advantages of utilizing an ideal in rough set theory are reducing the boundary region and improving the accuracy degree. Accordingly, the study of this theory with ideals is an interesting subject that has delivered the attention of many researchers [25–27]. As an important base for modulation of knowledge extraction and processing, this work will combine the notions of minimal structure approximation spaces and ideals (𝒥MSAS), where these concepts are not only in most fields of mathematics but also in several real-life problems.
The main contributions of this work are constructing ideal minimal structure approximation space 𝒥MSAS from minimal structure approximation space using the notion of ideal and applying them in the decision-making of the classification of amino acids. In addition, some sorts of near open and closed sets via 𝒥MSAS view are proposed and verified.
Preliminaries
Herein, some vital concepts and results are introduced, which are helpful in the sequel.
Definition 1 [13] Consider the binary relation η on the nonempty finite set Λ (named a universe set). Thus, the pair (Λ,η) is indicated as a generalized approximation space (briefly, an approximation space).
Definition 2 [13] Suppose that (Λ,η) is an approximation space. Thus, the right neighborhood Nr(s) of a point s∈Λ is defined by Nr(s)={y∈Λ:sηy}. Moreover, the class of all right neighborhoods in (Λ,η) is given by Nr(Λ)={Nr(s):s∈Λ}.
Definition 3 [22,23] For any approximation space (Λ,η), then:
(i) the minimal structure of (Λ,η) is the class (Λ)={∅,Λ}∪Nr(Λ). Accordingly, the triple (Λ,η,MS(Λ)) is titled a minimal structure approximation space (briefly, MSAS).
(ii) the elements of MS(Λ) are named MS(Λ)-open sets and their complements are MS(Λ)-closed sets.
(iii) the class of all MS(Λ)-closed sets, symbolized by (MS(Λ))c, is proposed by:
Definition 4 [22,23] Let B be a subset of an MSAS(Λ,η,MS(Λ)).Then:
(i) the minimal lower approximation of B is LMS(B)=∪{U∈MS(Λ):U⊆B}.
(ii) the minimal upper approximation of B is UMS(B)=∩{V∈(MS(Λ))c:B⊆V}.
(iii) the minimal accuracy of B is σr(B)=|LMS(B)||UMS(B)|, where |UMS(B)|≠0.
Proposition 1 [22] If η is a dominance (reflexive and transitive) relation on a universe Λ, therefore MS(Λ) represents a base for a topology on Λ.
Definition 5 [23] A subset B of an Λ is called:
(i) MS-regular open, if B = LMS(UMS(B)) (resp. MS-regular closed if B = UMS(LMS(B))).
(ii) MS-semi open, if B⊆UMS(LMS(B)) (resp. MS-semi closed if LMS(UMS(B))⊆B).
(iii) MS-pre open, if B⊆LMS(UMS(B)) (resp. MS-pre closed if UMS(LMS(B))⊆B).
The sets of all MS-regular open, MS-regular closed, MS-semi open, MS-semi closed, MS-pre open and MS-pre closed sets of (Λ,η,MS(Λ)) are symbolized as MS-RO(Λ), MS-RC(Λ), MS-SO(Λ), MS-SC(Λ), MS-PO(Λ) and MS-PC(Λ), respectively.
Generalized Rough Approximations Relying on Minimal Structure and Ideals
This section aims to introduce the ideal minimal structure approximation space 𝒥MSAS as a generalization of minimal structure approximation space MSAS via ideal. Some characteristics of the proposed method are obtained. Furthermore, the relations between the new approach 𝒥MSAS and the previous ones in [6,22,23] are studied.
Definition 6 Let (Λ,η,MS(Λ),𝒥) be an 𝒥MSAS. If B⊆Λ, then:
(i) a minimal lower approximation LMS𝒥(B) of B with respect to an ideal 𝒥 is
LMS𝒥(B)=η_r𝒥(B)∩B,whereη_r𝒥(B)=∪{U∈MS(Λ):U−B∈𝒥}.
(ii) a minimal upper approximation UMS𝒥(B) of B with respect to an ideal 𝒥 is
(i) the inversion of statement (iii) from Proposition 2 is not true. Suppose𝒥={∅,{b}}, and letB={b}, B`={a}. Accordingly, η_r𝒥(B)=∅, η_r𝒥(B`)={a,b}, η¯r𝒥(B)=∅, η¯r𝒥(B`)={a,b}. Hence, η_r𝒥(B)⊆η_r𝒥(B`)andη¯r𝒥(B)⊆η¯r𝒥(B`)althoughB⊈B`.
(ii) the inclusion of the first part of the statement (iv) of Proposition 2 cannot be exchanged by an equality symbol. Suppose𝒥={∅,{b}}, and letB={a}, B`={c}. Consequently, η_r𝒥(B)={a,b}, η_r𝒥(B`)={b,c}, andη_r𝒥(B∩B`)=∅. Hence, η_r𝒥(B)∩η_r𝒥(B`)⊈η_r𝒥(B∩B`).
(iii) the inclusion of the second part of the statement (iv) of Proposition 2 cannot be exchanged by equality symbol. Suppose𝒥={∅,{b}}, and letB={a,b,d}, B`={b,c,d}. Thus, η¯r𝒥(B)={a,d}, η¯r𝒥(B`) = {c,d}, andη¯r𝒥(B∪B`)=Λ. Hence, η¯r𝒥(B∪B`)⊈η¯r𝒥(B)∪η¯r𝒥(B`).
(iv) the inclusion of the first part of the statement (v) of Proposition 2 cannot be exchanged by equality symbol. Consider𝒥={∅,{c}}, and letB={a}, B`={b}. Therefore, η_r𝒥(B)={c}, η_r𝒥(B`)={b,c}, andη_r𝒥(B∪B`) = {a,b,c}. So, η_r𝒥(B∪B`)⊈η_r𝒥(B)∪η_r𝒥(B`).
(v) the inclusion of the second part of the statement (v) of Proposition 2 cannot be exchanged by equality symbol. Consider𝒥 = {∅,{c}}, and letB={a,c,d}, B`={b,c,d}. As a result, η¯r𝒥(B)={a,d}, η¯r𝒥(B`)={a,b,d}, andη¯r𝒥(B∩B`)={d}. Hence, η¯r𝒥(B∩B`)⊉η¯r𝒥(B)∩η¯r𝒥(B`).
The succeeding proposition is realized depending on Proposition 1.
Proposition 3 Let η, and MS(Λ) be a dominance relation, and a minimal structure on Λ, respectively. If B,B`⊆Λ, then for any ideal 𝒥 the following results hold: η_r𝒥(B∩B`) = η_r𝒥(B)∩η_r𝒥(B`), and η¯r𝒥(B∪B`) = η¯r𝒥(B)∪η¯r𝒥(B`).
The next remark is devoted to clarifying the differences between the existing approximations and the preceding one of Propositions 2 & 3 in [22].
Remark 3LetB, B`be subsets of an𝒥MSAS(Λ,η,MS(Λ),𝒥). Then Example 1 with𝒥={∅,{c}}confirms the following:
(i)η_r𝒥(∅)≠∅. Asη_r𝒥(∅)={c}.
(ii)η¯r𝒥(Λ)≠Λ. Asη¯r𝒥(Λ)={a,b,d}.
(iii)η_r𝒥(B)⊈B. AsB={a}, η_r𝒥(B)={c}.
(iv)B⊈η¯r𝒥(B). AsB={b,c,d}, η¯r𝒥(B)={a,b,d}.
Proposition 4 Let 𝒥,𝒥∙ be two ideals on an𝒥MSAS(Λ,η,MS(Λ)), and B⊆Λ. Then, the next results hold:
(i) if 𝒥⊆𝒥∙, then η_r𝒥(B)⊆η_r𝒥∙(B).
(ii) if 𝒥⊆𝒥∙, then η¯r𝒥∙(B)⊆η¯r𝒥(B).
(iii) η¯r(𝒥∪𝒥∙)(B)=η¯r𝒥(B)∪η¯r𝒥∙(B) and η¯r(𝒥∩𝒥∙)(B)=η¯r𝒥(B)∩η¯r𝒥∙(B).
(iv) η_r𝒥(B)∪η_r𝒥∙(B)=η_r(𝒥∪𝒥∙)(B) and η_r(𝒥∩𝒥∙)(B) = η_r𝒥(B)∩η_r𝒥∙(B).
(v) B∈𝒥 iff η¯r𝒥(B)=∅.
(vi) Bc∈𝒥 iff η_r𝒥(B)=Λ.
Proof. The first item will be proved, and the others similarly.
Let w∈η_r𝒥(B), and then there exists U∈MS(Λ) such that w∈U and U−B∈𝒥. Since 𝒥⊆𝒥∙, then w∈η_r𝒥∙(B) and so η_r𝒥(B)⊆η_r𝒥∙(B).
Example 2Letηbe a binary relation onΛ={a,b,c}such thatNr(a)={a}, Nr(b)=Λ, andNr(c)={a,c}. ThenMS(Λ)={∅,Λ,{a},{a,c}}, andMS(Λ)c={∅,Λ,{b},{b,c}}.
Remark 4Suppose𝒥={∅,{a}}, and𝒥∙={∅,{b}}in Example 2. Then,
(i) the inversion of statement (i) from Proposition 4 is false. SupposeB={b}, thenη_r𝒥∙(B)=∅, η_r𝒥(B)={a}. Hence, η_r𝒥∙(B)⊆η_r𝒥(B), while𝒥∙⊈𝒥.
(ii) the inversion of statement (ii) from Proposition 4 is not true. SupposeB = {a,b}, thenη¯r𝒥∙(B)=Λ, η¯r𝒥(B) = {b}. Hence, η¯r𝒥(B)⊆η¯r𝒥∙(B), while𝒥∙⊈𝒥.
Consequently, for any relation, the new kind of rough approximations𝒥MSASis more accurate than the preceding ones [6,22,23].
Remark 5Table 1compare between lower approximations, upper approximations, and the accuracy andTable 2compare between boundary, internal edge, and external edge defined in Definitions 4 and 6 which are given by the relationηof Example 1 and ideal𝒥={∅,{c}}.
Comparison between lower approximations, upper approximations, and the accuracy defined in Definitions 4 and 6 which are given by the relation η of Example 1, and an ideal 𝒥={∅,{c}}.
B⊆Λ
A method of Definition 4
Current method of Definition 6
LMS(B)
UMS(B)
σr(B)
LMS𝒥(B)
UMS𝒥(B)
σr𝒥(B)
{a}
∅
{a}
0
∅
{a}
0
{b}
∅
{a,b}
0
{b}
{a,b}
0.5
{c}
{c}
{c}
1
{c}
{c}
1
{d}
{d}
{d}
1
{d}
{d}
1
{a,b}
{a,b}
{a,b}
1
{a,b}
{a,b}
1
{a,c}
{c}
{a,b,c}
0.33
{c}
{a,c}
0.5
{a,d}
{d}
{a,d}
0.5
{d}
{a,d}
0.5
{b,c}
{b,c}
{a,b,c}
0.67
{b,c}
{a,b,c}
0.67
{b,d}
{d}
{a,b,d}
0.33
{b,d}
{a,b,d}
0.67
{c,d}
{c,d}
{c,d}
1
{c,d}
{c,d}
1
{a,b,c}
{a,b,c}
{a,b,c}
1
{a,b,c}
{a,b,c}
1
{a,b,d}
{a,b,d}
{a,b,d}
1
{a,b,d}
{a,b,d}
1
{a,c,d}
{c,d}
Λ
0.5
{c,d}
{a,c,d}
0.67
{b,c,d}
{b,c,d}
Λ
0.75
{b,c,d}
Λ
0.75
Λ
Λ
Λ
1
Λ
Λ
1
Comparison between the boundary, internal edge, and external edge defined in Definitions 4 and 6 which are given by the binary relation η of Example 1, and an ideal 𝒥={∅,{c}}.
B⊆Λ
A method of Definition 4
Current method of Definition 6
MS-br(B)
LMS-edg(B)
UMS-edg(B)
MS-br𝒥(B)
LMS𝒥-edg(B)
UMS𝒥-edg(B)
{a}
{a}
{a}
∅
{a}
{a}
∅
{b}
{a, b}
{b}
{a}
{a}
∅
{a}
{c}
∅
∅
∅
∅
∅
∅
{d}
∅
∅
∅
∅
∅
∅
{a, b}
∅
∅
∅
∅
∅
∅
{a, c}
{a, b}
{a}
{b}
{a}
{a}
∅
{a, d}
{a}
{a}
∅
{a}
{a}
∅
{b, c}
{a}
∅
{a}
{a}
∅
{a}
{b, d}
{a, b}
{b}
{a}
{a}
∅
{a}
{c, d}
∅
∅
∅
∅
∅
∅
{a, b, c}
∅
∅
∅
∅
∅
∅
{a, b, d}
∅
∅
∅
∅
∅
∅
{a, c, d}
{a, b}
{a}
{b}
{a}
{a}
∅
{b, c, d}
{a}
∅
{a}
{a}
∅
{a}
Λ
∅
∅
∅
∅
∅
∅
The next result exhibits the connections among the lower, and upper approximations and the degree of accuracy that were offered in both Definitions 4, and 6.
Theorem 1 Let (Λ,η,MS(Λ),𝒥) be an 𝒥MSAS. If B⊆Λ. Then, the next properties are held:
(i) LMS(B)⊆LMS𝒥(B).
(ii) UMS𝒥(B)⊆UMS(B).
(iii) σr(B)<σr𝒥(B).
Proof. The first item will be proved, and the others similarly.
Let w∈LMS(B), and then there exists U∈MS(Λ) such that w∈U and U⊆B. Hence, U−B=∅∈𝒥. i.e., w∈η_r𝒥(B) . So, LMS(B)⊆η_r𝒥(B). Since LMS(B)⊆B, therefore LMS(B)⊆LMS𝒥(B).
Remark 6It must be perceived that the suggested approximationsLMS𝒥(.), andUMS𝒥(.)in Definition 6 have equal characteristics ofη_r𝒥(.)andη¯r𝒥(.)identified in Propositions 1 and 3. Moreover, it fulfills the next properties:
(i) LMS𝒥(∅)=∅andUMS𝒥(Λ)=Λ.
(ii) LMS𝒥(B)⊆B⊆UMS𝒥(B).
Definition 7 For an 𝒥MSAS(Λ,η,MS(Λ),𝒥). If B⊆Λ, then
(i) a minimal positive of B with respect to an ideal 𝒥 is MS-Por𝒥(B)=LMS𝒥(B),
(ii) a minimal exterior (negative) of B with respect to an ideal 𝒥 is MS-Exr𝒥(B)=Λ−UMS𝒥(B),
(iii) a minimal boundary of B with respect to ideal 𝒥 is MS-br𝒥(B)=UMS𝒥(B)−LMS𝒥(B),
(iv) a minimal internal edge of B with respect to an ideal 𝒥 is LMS𝒥-edg(B)=B−LMS𝒥(B), and
(v) a minimal external edge of B with respect to an ideal 𝒥 is UMS𝒥-edg(B)=UMS𝒥(B)−B.
Remark 7 According to Example 1 with 𝒥={∅,{c}}, Table 2 illustrates the next relationships for B⊆Λ:
(i) MS-br𝒥(B)⊆MS-br(B),
(ii) LMS𝒥-edg(B)⊆LMS-edg(B), and
(iii) UMS𝒥-edg(B)⊆UMS-edg(B).
Definition 8 For an 𝒥MSAS(Λ,η,MS(Λ),𝒥), a subset B of Λ is categorized as:
(i) MS-definable (exact) with respect to an ideal 𝒥, if MS-br𝒥(B)=∅. Otherwise, a subset B is called MS-undefinable (rough) with respect to the 𝒥.
(ii) MS-internally definable with respect to an ideal 𝒥, if LMS𝒥(B)=B.
(iii) MS-externally definable with respect to an ideal 𝒥, if UMS𝒥(B)=B.
Remark 8According toTable 1and Example 1 with𝒥={∅,{c}}, it is noticed that:
(i) the sets{c},{d},{a,b},{c,d},{a,b,c},{a,b,d}, andΛareMS-definable with respect to an ideal𝒥.
(ii) the sets{b},{b,c},{b,d}, and{b,c,d}areMS-internally definable with respect to an ideal𝒥.
(iii) the sets{a},{a,c},{a,d}, and{a,c,d}areMS-externally definable with respect to an ideal𝒥.
Corollary 1For a𝒥MSAS(Λ,η,MS(Λ),𝒥). Then,
(i) every MS-definable is MS-definable with respect to an ideal 𝒥.
(ii) every MS-undefinable with respect to an ideal 𝒥 is MS-undefinable.
Some Near Open Sets Based on MSAS and Ideals
This section aims to introduce and investigate some sorts of near open (resp. closed) sets via the viewpoint of an 𝒥MSAS.
Definition 9 A subset B of an 𝒥MSAS(Λ,η,MS(Λ),𝒥) is called:
(i) MS𝒥-regular open if B = LMS𝒥(UMS𝒥(B))
(ii) MS𝒥-semi open if B⊆UMS𝒥(LMS𝒥(B)).
(iii) MS𝒥-pre open if B⊆LMS𝒥(UMS𝒥(B)).
Remark 9
(i) The complement of anMS𝒥-regular open (resp. MS𝒥-semi open, andMS𝒥-pre open) set is knownMS𝒥-regular closed (resp. MS𝒥-semi closed, andMS𝒥-pre closed) set.
(ii) The set of allMS𝒥-regular open (resp. MS𝒥-regular closed, MS𝒥-semi open, MS𝒥-semi closed, MS𝒥-pre open andMS𝒥-pre closed) sets of(Λ,η,MS(Λ))is denoted byMS𝒥-RO(Λ)(resp. MS𝒥-RC(Λ), MS𝒥-SO(Λ), MS𝒥-SC(Λ), MS𝒥-PO(Λ)andMS𝒥-PC(Λ)).
Remark 10In a topological space, the class of semi (resp. pre-) open sets is contained in the class of ideal semi (resp. pre-) open sets. While this fact is discussed for minimal structures, surprisingly it is incorrect i.e., MS-SO(Λ)andMS𝒥-SO(Λ)(resp. MS-PO(Λ)andMS𝒥-PO(Λ)) are incomparable, as shown in Examples 3, and 4.
Example 3 (Continued from Example 2) Consider𝒥={∅,{a},{b},{a,b}}is an ideal on(Λ,η,MS(Λ)), that will result in.
(i) MS-RO(Λ)={∅,Λ}.
(ii) MS-SO(Λ)={∅,{a},{a,c},{a,b},Λ}.
(iii) MS-PO(Λ)={∅,{a},{a,c},{a,b},Λ}.
(iv) MS𝒥-RO(Λ)={∅,{a},{b,c},Λ}.
(v) MS𝒥-SO(Λ)={∅,{a},{c},{a,c},{b,c},Λ}.
(vi) MS𝒥-PO(Λ)={∅,{a},{c},{a,c},{b,c},Λ}.
Example 4ConsiderΛ={a,b,c,d}, ηis a binary relation onΛ, andNr(a)={a}, Nr(b)={a,b}, Nr(c)={c}, andNr(d)={b,c,d}. Accordingly, MS(Λ)={∅,Λ,{a},{c},{b,c,d},{a,b}}and
MS(Λ)c={∅,Λ,{a},{c,d},{a,b,d},{b,c,d}}. If𝒥={∅,{a},{d},{a,d}}, andL={∅,{b},{d},{b,d}}be ideals on(Λ,η,MS(Λ)), then
(i) MS-RO(Λ)={∅,Λ,{a},{c},{b,c,d},{a,b}}.
(ii) MS-SO(Λ)={∅,Λ,{a},{c},{a,b},{a,c},{c,d},{a,b,c},{a,b,d},{b,c,d}}.
According to Definition 9, the proof of the proposition 5 is obvious.
Proposition 5 Let (Λ,η,MS(Λ),𝒥) be an 𝒥MSAS, then MS𝒥-RO(Λ)⊆MS𝒥-SO(Λ)∩MS𝒥-PO(Λ).
The next example shows that the converse of Proposition 5 is incorrect.
Example 5In Example 4, {b,c}is anMS𝒥-pre open set and it is notMS𝒥-regular open. In addition, {b,d}isMS𝒥-semi open and it is notMS𝒥-regular open.
Remark 11In view of Example 4, the following results are noticed:
(i) the union ofMS𝒥-regular open sets is notMS𝒥-regular open. Consider{b},{c}∈MS𝒥-RO(Λ). Clearly, {b,c}∉MS𝒥-RO(Λ).
(ii) the intersection ofMS𝒥-regular open sets is notMS𝒥-regular open. Consider{a,b},{b,c,d}∈MS𝒥-RO(Λ). Clearly, {b}∉MS𝒥-RO(Λ).
(iii) the intersection ofMS𝒥-semi open sets is notMS𝒥-semi open. Consider{a,b,d},{a,c,d}∈MS𝒥-SO(Λ). Clearly, {a,d}∉MS𝒥-SO(Λ).
(iv) the intersection ofMS𝒥-pre open sets is notMS𝒥-pre open. Consider{a,b},{b,c}∈MS𝒥-PO(Λ). Clearly, {b}∉MS𝒥-PO(Λ).
(v) the intersection ofMS𝒥-regular open set andMS𝒥-pre open set is notMS𝒥-pre open set. Consider{a,b}∈MS𝒥-RO(Λ), and{b,c,d}∈MS𝒥-PO(Λ). Clearly, {b}∉MS𝒥-PO(Λ).
(vi) the intersection ofMS𝒥-regular open set andMS𝒥-semi open set is notMS𝒥-semi open set. Consider{a,b}∈MS𝒥-RO(Λ), and{b,c,d}∈MS𝒥-SO(Λ). Clearly, {b}∉MS𝒥-SO(Λ).
Biochemical Applications
An example in the area of chemistry is provided by utilizing the actual approximation in Definition 6 to clarify the notions practically.
Example 6 [26] LetΛ={m1,m2,m3,m4,m5}be five amino acids (AAs). The (AAs) is qualified by the attributesA1,A2,A3,A4,A5such thatA1refers to PIE, A2refers to SAC (surface area), A3refers to MR (molecular refractivity), A4refers to LAM (the side chain polarity), andA5refers to Vol (molecular volume). Table 3shows all quantitative attributes of AAs.
Quantitative attributes of AAs
A1
A2
A3
A4
A5
m1
0.23
254.2
2.126
−0.02
82.2
m2
−0.48
303.6
2.994
−1.24
112.3
m3
−0.61
287.9
2.994
−1.08
103.7
m4
0.45
282.9
2.933
−0.11
99.1
m5
−0.11
335.0
3.458
−0.19
127.5
Presently, it shall be investigated five relations on Λ determined by ηl={(mi,mj)∈Λ×Λ:mi(Al)−mj(Al)<δl2,i,j,l=1,2,3,4,5}, where δl symbolizes the standard deviation of the quantitative attributes Al,l=1,2,3,4,5. Thus, the right neighborhoods for all members of Λ according to the relations ηl,l=1,2,3,4,5 are tabulated in Table 4.
Right neighborhoods of five relations
mlη1
mlη2
mlη3
mlη4
mlη15
m1
{m1,m4}
Λ
Λ
{m1,m4,m5}
Λ
m2
Λ
{m2,m5}
{m2,m3,m4,m5}
Λ
{m2,m5}
m3
Λ
{m2,m3,m4,m5}
{m2,m3,m4,m5}
Λ
{m2,m3,m4,m5}
m4
{m4}
{m2,m3,m4,m5}
{m2,m3,m4,m5}
{m1,m4,m5}
{m2,m3,m4,m5}
m5
{m1,m4,m5}
{m5}
{m5}
{m1,m4,m5}
{m5}
The intersection of all right neighborhoods of all elements l=1,2,3,4,5 is computed as follows: mlη=∩l=15mlη1={m1,m4}, m2η=∩l=15m2η1={m2,m5}, m3η=∩l=15m3η1={m2,m3,m4,m5}, m4η=∩l=15m4η1={m4}, and m5η=∩l=15m5η1={m5}. Then, it will be obtained MS(Λ) = {∅,Λ,{m4},{m5},{m1,m4},{m2,m5},{m2,m3,m4,m5}}.
The ideal minimal accuracy measure σr𝒥(.) that is calculated by using the current approximation in Definition 6 increased more than the minimal accuracy measure σr(.) due to Definition 4, for any subset of Λ as tabulated in Table 5.
An Algorithm and Framework
This section provides an algorithm and a framework for decision-making problems. The suggested algorithm is checked with fictitious data and compared to existing methods. This technique represents a simple tool that can be used in MATLAB.
Require: Initiate an information table generated from the given data such that the first column contains a set of objects Λ, and the set of attributes as a first row.
Output: An accurate decision for exact and rough sets.
Step 1: Input a finite set of data as a universal set Λ, and Al a set of attributes from the information table.
Step 2: Define the binary relations ηl={(mi,mj)∈Λ×Λ:mi(Al)−mj(Al)<δl2,i,j,l=1,2,3,4,5}.
Step 3: Compute all right neighborhoods of all elements by miη=∩l=15miη1, for each i,l=1,2,3,4,5
Step 4: Construct the class of minimal structure by Step 3.
Step 5: Using the ideal 𝒥 (which is given by an expert), compute LMS𝒥(B) and UMS𝒥(B) of B⊆Λ with respect to ideal 𝒥 as follows: LMS𝒥(B) = η_r𝒥(B)∩B, where η_r𝒥(B) = ∪{U∈MS(Λ):U−B∈𝒥}, and UMS𝒥(B) = η¯r𝒥(B)∪B, where η¯r𝒥(B) = ∩{V∈(MS(Λ))c:B−V∈𝒥}.
Step 6: Using the ideal 𝒥 (which is given by an expert), compute the minimal accuracy of the approximations in Step 5 of all subsets in Λ by σr𝒥(B) = |LMS𝒥(B)||UMS𝒥(B)|, where |UMS𝒥(B)|≠0.
Step 7: If σr𝒥(B)=1, then B is an exact set. Else, B is a rough set.
The following figure (Fig. 1) illustrates a simple flowchart for calculating the degree of accuracy induced from the above algorithm.
A flowchart for decision making using an 𝒥MSAS
Conclusions and Discussions
The novel rough approximation space 𝒥MSAS generated by the minimal structure and ideal concepts were proposed, and their principal characteristics were verified. The best approximations and degrees of accuracy have been achieved. A novel approach has been compared with the other approaches in the references [6,22,23] via counterexamples that has been examined as indicated in Theorem 1 and Table 1. From Remark 3, some principal properties of rough sets concerning η_r𝒥(), and η¯r𝒥() were deduced. By increasing the number of elements of ideals, the lower approximation would increase and the upper approximation would decrease and hence the measure of accuracy becomes more accurate as given in Proposition 4. In addition, the minimal internal edge, minimal external edge, and the degree of accuracy were described by using minimal boundary. Several sorts of near open and near closed sets by the 𝒥MSAS view were studied.
One of the challenges in daily problems, as in the medical diagnosis, is making an accurate decision. Therefore, the applied example in biochemistry offers a clear vision that the expansion using the ideal gives better results. Thus, by the 𝒥MSAS different sorts of mathematical tools, which may help experts in studying amino acids, were suggested. A simple approach is used in MATLAB, and the proposed method in an algorithm form was demonstrated. In reality, that approach may be useful in solving some future real-life problems.
In the forthcoming, the 𝒥MSAS approach will be extended to a variety of other concepts, such as fuzzy sets and soft rough sets. Also, it is planned to benefit from the 𝒥MSAS approach to apply them to the problems in [4,8,14,15] to improve their accuracy values.
We appreciate the reviewers for their invaluable time in reviewing our paper and providing thoughtful and valuable comments. It was their insightful suggestions that led to sensible improvements in the current version.
Funding Statement
The authors received no specific funding for this study.
Conflicts of Interest
The authors declare that they have no conflicts of interest to report regarding the present study.
ReferencesA.Galton, “A generalized topological view of motion in discrete space,” , vol. 305, no. 1–3, pp. 111–134, 2003.B.Stadler and P.Stadler, “Generalized topological spaces in evolutionary theory and combinatorial chemistry,” , vol. 42, no. 3, pp. 577–585, 2002; 12086517R.Abu-Gdairi, M. A.El-Gayar, T. M.Al-shami, A. S.Nawar and M. K.El-Bably, “Some topological approaches for generalized rough sets and their decision-making applications,” , vol. 14, no. 1, pp. 95–119, 2022.M. K.El-Bably, M. I.Ali and E. A.Abo-Tabl, “New topological approaches to generalized soft rough approximations with medical applications,” , vol. 2021, no. 2559495, pp. 1–16, 2021.E. F.Lashin, A. M.Kozae, A. A.Abo Khadra and T.Medhat, “Rough set theory for topological spaces,” , vol. 40, no. 1–2, pp. 35–43, 2005.Z.Pawlak, “Rough sets,” , vol. 11, no. 5, pp. 341–356, 1982.M.Abdelaziz, H. M.Abu-Donia, R. A.Hosny, S. L.Hazae and R. A.Ibrahim, “Improved evolutionary-based feature selection technique using extension of knowledge based on the rough approximations,” , vol. 594, pp. 76–94, 2022.M. A.El-Gayar and A. A. E.Atik, “Topological models of rough sets and decision making of COVID-19,” , vol. 2022, no. 2989236, pp. 1– 10, 2022.R.Abu-Gdairi, M. A.El-Gayar, M. K.El-Bably and K. K.Fleifel, “Two different views for generalized rough sets with applications,” , vol. 9, no. 18, pp. 2275–2296, 2021.R. A.Hosny, M.Abdelaziz and R. A.Ibrahim, “Enhanced feature selection based on integration containment neighborhoods rough set approximations and binary honey badger optimization,” , vol. 2022, no. 3991870, pp. 1–17, 2022.K.Qin, J.Yang and Z.Pei, “Generalized rough sets based on reflexive and transitive relations,” , vol. 178, no. 21, pp. 4138–4141, 2008.M.Kondo, “On structure of generalized rough sets,” , vol. 176, no. 5, pp. 589–600, 2006.Y. Y.Yao, “Two views of the theory of rough sets in finite universes,” , vol. 15, no. 4, pp. 291–317, 1996.M. K.El-Bably and M. E.Sayed, “Three methods to generalize Pawlak approximations via simply open concepts with economic applications,” , vol. 26, no. 10, pp. 4685–4700, 2022.M. E.Abd El-Monsef, M. A.EL-Gayar and R. M.Aqeel, “On relationships between revised rough fuzzy approximation operators and fuzzy topological spaces,” , vol. 3, no. 4, pp. 257–271, 2014.M. E.Abd El-Monsef, M. A.El-Gayar and R. M.Aqeel, “A comparison of three types of rough fuzzy sets based on two universal sets,” , vol. 8, no. 1, pp. 343–353, 2017.A. T.Azar, M. S.Elgendy, M.Abdul Salam and K. M.Fouad, “Rough sets hybridization with mayfly optimization for dimensionality reduction,” , vol. 73, no. 1, pp. 1087–1108, 2022.A. B.Khoshaim, S.Abdullah, S.Ashraf and M.Naeem, “Emergency decision-making based on q-rung orthopair fuzzy rough aggregation information,” , vol. 69, no. 3, pp. 4077–4094, 2021.X.Li, S.Zhou, Z.An and Z.Du, “Multi-span and multiple relevant time series prediction based on neighborhood rough set,” , vol. 67, no. 3, pp. 3765–3780, 2021.V.Popa and T.Noiri, “On the definitions of some generalized forms of continuity under minimal conditions,” , vol. 22, pp. 9–18, 2001.M.Alimohammady and M.Roohi, “Linear minimal space,” , vol. 33, no. 4, pp. 1348–1354, 2007.M. M.EL-Sharkasy, “Minimal structure approximation space and some of its application,” , vol. 40, no. 1, pp. 973–982, 2021.M. M.EL-Sharkasy and F.Altwme, “Some new types of approximations via minimal structure,” , vol. 26, no. 1, pp. 287–296, 2018.K.Kuratowski, , vol. 1. New York: Academic Press, 1966.R. A.Hosny, T. M.Al-shami, A. A.Azzam and A. S.Nawar, “Knowledge based on rough approximations and ideals,” , vol. 2022, no. 3766286, pp. 1–12, 2022.A. S.Nawar, M. A.El-Gayar, M. K.El-Bably and A. R.Hosny, “θβ-ideal approximation spaces and their applications,” , vol. 7, no. 2, pp. 2479–2497, 2022.R. A.Hosny, B. A.Asaad, A. A.Azzam and T. M.Al-shami, “Various topologies generated from Ej-neighbourhoods via ideals,” , vol. 2021, no. 4149368, pp. 1–11, 2021.