Type-2 and Type-𝑛 Rough Sets: A Hierarchical Generalization of Pawlak
Approximations
Takaaki Fujita 1 ∗
1 Independent Researcher, Shinjuku, Shinjuku-ku, Tokyo, Japan.
Abstract
Uncertainty is ubiquitous in real-world systems and has stimulated the development of many generalized set-
theoretic models, including rough sets, fuzzy sets, intuitionistic fuzzy sets, and neutrosophic sets. Rough
set theory, in particular, represents uncertainty by approximating a target subset of a universe through two
crisp sets—the lower and upper approximations—constructed from an equivalence (indiscernibility) relation.
Motivated by the well-established hierarchy of Type-2 and, more generally, Type-𝑛 fuzzy sets, this paper
proposes and studies analogous higher-order extensions of rough sets. We introduce Type-2 rough sets as
a two-level parameterization that assigns, to each primary context, a family of Pawlak approximation pairs
under secondary parameters, and we then generalize this idea recursively to define Type-𝑛 rough sets. These
hierarchical rough frameworks provide a principled mechanism for modeling context-dependent granulations
and multi-level uncertainty within a unified approximation-based formalism.
Keywords: Rough Set, Type-2 Rough Set, Type-n Rough Set
1 Introduction
1.1 Rough Set Theory
Uncertainty is ubiquitous in real-world systems, and its mathematical treatment has led to a broad spectrum
of generalized set-theoretic frameworks, including rough sets, fuzzy sets [1], intuitionistic fuzzy sets [2],
neutrosophic sets [3], soft sets [4], and plithogenic sets [5]. These theories have been extensively studied and
have found applications across many domains.
In this paper, we focus on rough sets. Rough set theory provides a systematic approach to uncertainty by
approximating a target subset of a universe through two crisp sets—the lower and upper approximations—
constructed from an equivalence (indiscernibility) relation on the universe [6]. To accommodate more complex
decision environments, numerous extensions of rough sets have been proposed, including TreeRough sets [7],
weighted rough sets [8, 9], fuzzy rough sets [10], game-theoretic rough sets [11], probabilistic rough sets [12],
and dominance-based rough sets [13]. These variants, like the classical model, have been actively investigated
in decision science and many other fields, underscoring the continuing importance of rough set theory.
1.2 Our Contributions
Higher-order paradigms are well established in fuzzy set theory, most notably Type-2 and, more generally,
Type-𝑛 fuzzy sets [14, 15]. Closely related hierarchical extensions have also been developed for neutrosophic
sets [16, 17] and for soft sets [18], together with a wide range of applications. In contrast, analogous higher-
order generalizations for rough sets have received comparatively limited attention. To help bridge this gap, and
inspired by the hierarchy of Type-2 and Type-𝑛 fuzzy sets, we investigate corresponding extensions of rough
sets, namely Type-2 and Type-𝑛 rough sets. For reference, Table 1 provides a concise comparison of rough
sets, Type-2 rough sets, and Type-𝑛 rough sets.
1.3 Structure of this paper
The format of this paper is described below.
1 Introduction 1
1.1 Rough Set Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Our Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Structure of this paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1
Table 1: Concise comparison of Rough Sets, Type-2 Rough Sets, and Type-𝑛 Rough Sets.
Aspect Rough set (Pawlak) Type-2 rough set (T2RS) Type-𝑛 rough set (T𝑛RS)
Universe / data do- Finite universe 𝑈. Finite universe 𝑈 (same). Finite universe 𝑈 (same).
main
Granulation / relation Single equivalence relation 𝑅 ∈ Parameterized space (𝑈, 𝐸, 𝜌) Same parameterized space
input Eq(𝑈). with 𝜌 : 𝐸 → Eq(𝑈); different (𝑈, 𝐸, 𝜌), used recursively across
parameters yield different 𝑅’s. levels.
Object being approx- A subset 𝑋 ⊆ 𝑈. Typically a family of targets Targets may vary along
imated (𝑋 𝛼 ) 𝛼∈ 𝐴, one per primary param- branches/levels; ultimately
eter 𝛼 (or a common 𝑋 if im- each leaf corresponds to some 𝑋
posed). and some 𝑒 ∈ 𝐸.
Primary parameter None. Nonempty 𝐴 ⊆ 𝐸 (primary pa- Nonempty 𝐴𝑛 ⊆ 𝐸 at level 𝑛
set rameters). (and implicitly lower-level param-
eter sets via sections).
Approximation out- Single pair (𝑋, 𝑋). Map 𝛼 ↦→ a type-1 rough set (a Recursive map 𝑎 ↦→ type-(𝑛 − 1)
𝑒
put family 𝑒 ↦→ (𝑋 𝛼 𝑒 , 𝑋 𝛼 ) over 𝐿 𝛼 ). rough set; leaf evaluations yield
Pawlak pairs.
Structural interpreta- One granulation, one “defi- Two-level choice: primary context Hierarchical contexts (policy →
tion nitely/possibly” boundary. 𝛼 selects a bundle of granulations viewpoint → protocol → · · · )
𝐿 𝛼. selecting lower-level rough struc-
tures.
Reduction to Pawlak Trivial (it is Pawlak). If | 𝐴| = |𝐿 𝛼 | = 1, reduces to one If the hierarchy collapses to a sin-
case Pawlak pair. gle path with singleton choices at
each level, reduces to one Pawlak
pair.
2 Preliminaries 2
2.1 Rough Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3 Results 3
3.1 Type-2 Rough Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.2 Type-n Rough Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4 Conclusion 7
2 Preliminaries
In this section, we briefly review the definitions and basic concepts used throughout the paper.
2.1 Rough Set
A rough set approximates a target subset using lower and upper sets induced by an indiscernibility (equivalence)
relation on the universe.
Definition 2.1 (Rough Set Approximation). [19] [6, 20] Let 𝑋 be a finite universe and let
𝑅 ⊆ 𝑋×𝑋
be an equivalence relation, whose equivalence classes are written [𝑥] 𝑅 for each 𝑥 ∈ 𝑋. For any subset 𝑌 ⊆ 𝑋,
define:
𝑌 = 𝑥 ∈ 𝑋 | [𝑥] 𝑅 ⊆ 𝑌 , 𝑌 = 𝑥 ∈ 𝑋 | [𝑥] 𝑅 ∩ 𝑌 ≠ ∅ .
Here 𝑌 collects all elements whose entire indiscernibility class lies inside 𝑌 (those that definitely belong), while
𝑌 gathers elements whose class meets 𝑌 nontrivially (those that possibly belong). The pair (𝑌 , 𝑌 ) is called the
rough approximation of 𝑌 , and satisfies
𝑌 ⊆ 𝑌 ⊆ 𝑌.
2
3 Results
In this section, we present the main results of the paper.
3.1 Type-2 Rough Set
A Type-2 Rough Set is a two-level rough framework in which each primary parameter selects a rough approxi-
mation family under secondary parameters and equivalence relations.
Definition 3.1 (Parameterized approximation space). Let 𝑈 be a finite universe and let 𝐸 be a nonempty finite
set of parameters. A parameterized approximation space is a triple
PA = (𝑈, 𝐸, 𝜌),
where 𝜌 : 𝐸 → Eq(𝑈) assigns to each parameter 𝑒 ∈ 𝐸 an equivalence relation 𝜌(𝑒) ⊆ 𝑈 × 𝑈 on 𝑈. For each
𝑒 ∈ 𝐸 and 𝑥 ∈ 𝑈, write [𝑥] 𝑒 := [𝑥] 𝜌(𝑒) for the corresponding equivalence class.
Definition 3.2 (Type-1 rough set (parameterized Pawlak rough set)). Let PA = (𝑈, 𝐸, 𝜌) be a parameterized
approximation space and fix a nonempty subset 𝐵 ⊆ 𝐸. For a target set 𝑋 ⊆ 𝑈, define for each 𝑒 ∈ 𝐵 the
Pawlak lower and upper approximations
𝑒
𝑋 𝑒 := { 𝑥 ∈ 𝑈 | [𝑥] 𝑒 ⊆ 𝑋 }, 𝑋 := { 𝑥 ∈ 𝑈 | [𝑥] 𝑒 ∩ 𝑋 ≠ ∅ }.
The pair
RS (1) (𝑋, 𝐵) := (𝐹𝑋,𝐵 , 𝐵),
𝑒
𝐹𝑋,𝐵 : 𝐵 → P (𝑈) × P (𝑈), 𝐹𝑋,𝐵 (𝑒) := (𝑋 𝑒 , 𝑋 ),
is called the type-1 rough set of 𝑋 over 𝐵 (in PA).
Definition 3.3 (Type-2 rough set). Let PA = (𝑈, 𝐸, 𝜌) be a parameterized approximation space. Let
Σ (1) (𝑈, 𝐸, 𝜌) denote the collection of all type-1 rough sets RS (1) (𝑋, 𝐵) obtained from subsets 𝑋 ⊆ 𝑈
and nonempty 𝐵 ⊆ 𝐸.
A type-2 rough set (briefly, T2RS) over PA is a pair (F (2) , 𝐴) consisting of a nonempty primary parameter set
𝐴 ⊆ 𝐸 and a mapping
F (2) : 𝐴 → Σ (1) (𝑈, 𝐸, 𝜌), 𝛼 ↦−→ F (2) (𝛼) = (𝐹𝛼 , 𝐿 𝛼 ),
where for each 𝛼 ∈ 𝐴:
(i) 𝐿 𝛼 ⊆ 𝐸 is a nonempty underlying parameter set, and
(ii) there exists a target set 𝑋 𝛼 ⊆ 𝑈 such that
𝑒
𝐹𝛼 (𝑒) = 𝑋 𝛼 𝑒 , 𝑋 𝛼 for all 𝑒 ∈ 𝐿 𝛼 .
Ð
The set 𝛼∈ 𝐴 𝐿 𝛼 is called the underlying parameter set of the T2RS.
Remark 3.4. If 𝐴 = {𝛼} and 𝐿 𝛼 = {𝑒} are singletons, then a type-2 rough set reduces to the classical Pawlak
𝑒
rough approximation pair (𝑋 𝛼 𝑒 , 𝑋 𝛼 ).
Definition 3.5 (Underlying parameter set and sections). Let T = (F (2) , 𝐴) be a type-2 rough set as in
Definition 3.3, with F (2) (𝛼) = (𝐹𝛼 , 𝐿 𝛼 ).
(i) The underlying parameter set of T is
Ø
Und(T) := 𝐿𝛼 ⊆ 𝐸.
𝛼∈ 𝐴
(ii) For each 𝛼 ∈ 𝐴, the object T ↓ 𝛼 := F (2) (𝛼) = (𝐹𝛼 , 𝐿 𝛼 ) is called the section of T at 𝛼 (a type-1 rough
set).
3
Theorem 3.6 (Leaf roughness for type-2 rough sets). Let T = (F (2) , 𝐴) be a type-2 rough set over PA. For
every 𝛼 ∈ 𝐴 and every 𝑒 ∈ 𝐿 𝛼 , there exists a target set 𝑋 𝛼 ⊆ 𝑈 such that
𝑒 𝑒
𝐹𝛼 (𝑒) = 𝑋 𝛼 𝑒 , 𝑋 𝛼 and hence 𝑋 𝛼 𝑒 ⊆ 𝑋 𝛼 ⊆ 𝑋 𝛼 .
In particular,
𝑒
𝑋𝛼 𝑒 ⊆ 𝑋𝛼 .
𝑒
Proof. Fix 𝛼 ∈ 𝐴. By Definition 3.3(ii), there exists 𝑋 𝛼 ⊆ 𝑈 such that 𝐹𝛼 (𝑒) = (𝑋 𝛼 𝑒 , 𝑋 𝛼 ) for all 𝑒 ∈ 𝐿 𝛼 .
𝑒 𝑒
For any fixed 𝑒 ∈ 𝐿 𝛼 , the Pawlak approximation properties yield 𝑋 𝛼 𝑒 ⊆ 𝑋 𝛼 ⊆ 𝑋 𝛼 , whence 𝑋 𝛼 𝑒 ⊆ 𝑋 𝛼
follows immediately. □
Theorem 3.7 (Closure under restriction of primary parameters). Let T = (F (2) , 𝐴) be a type-2 rough set and
let 𝐴′ ⊆ 𝐴 be nonempty. Define the restriction
T 𝐴′ := (F (2) ↾ 𝐴′ , 𝐴′ ).
Then T 𝐴′ is again a type-2 rough set over PA. Moreover, for each 𝛼 ∈ 𝐴′ ,
T 𝐴′ ↓ 𝛼 = T ↓ 𝛼.
Proof. Since T is a type-2 rough set, for every 𝛼 ∈ 𝐴 we have F (2) (𝛼) ∈ Σ (1) (𝑈, 𝐸, 𝜌). Therefore, for
every 𝛼 ∈ 𝐴′ the restricted map F (2) ↾ 𝐴′ still satisfies F (2) ↾ 𝐴′ (𝛼) = F (2) (𝛼) ∈ Σ (1) (𝑈, 𝐸, 𝜌). Hence
(F (2) ↾ 𝐴′ , 𝐴′ ) satisfies Definition 3.3 and is a type-2 rough set. The identity of sections is immediate from the
definition of restriction. □
Theorem 3.8 (Closure under restriction of underlying parameters). Let T = (F (2) , 𝐴) be a type-2 rough set,
and for each 𝛼 ∈ 𝐴 let 𝐿 ′𝛼 ⊆ 𝐿 𝛼 be nonempty. Define a new mapping F ′(2) : 𝐴 → Σ (1) (𝑈, 𝐸, 𝜌) by
F ′(2) (𝛼) := 𝐹𝛼↾ 𝐿 ′𝛼 , 𝐿 ′𝛼 ,
𝛼 ∈ 𝐴,
and set T′ := (F ′(2) , 𝐴). Then T′ is a type-2 rough set over PA.
Proof. Fix 𝛼 ∈ 𝐴. Since F (2) (𝛼) = (𝐹𝛼 , 𝐿 𝛼 ) ∈ Σ (1) (𝑈, 𝐸, 𝜌), there exists a target set 𝑋 𝛼 ⊆ 𝑈 such that
𝑒
𝐹𝛼 (𝑒) = (𝑋 𝛼 𝑒 , 𝑋 𝛼 ) for all 𝑒 ∈ 𝐿 𝛼 . In particular, the same equality holds for all 𝑒 ∈ 𝐿 ′𝛼 ⊆ 𝐿 𝛼 . Hence the
restricted pair (𝐹𝛼 ↾ 𝐿 ′𝛼 , 𝐿 ′𝛼 ) is a type-1 rough set, i.e., belongs to Σ (1) (𝑈, 𝐸, 𝜌). Since this holds for every
𝛼 ∈ 𝐴, the mapping F ′(2) satisfies Definition 3.3, and thus T′ is a type-2 rough set. □
Definition 3.9 (Pointwise union and intersection under a common target). Let T = (F (2) , 𝐴) and S = (G (2) , 𝐴)
be type-2 rough sets over the same PA with the same primary parameter set 𝐴. Assume that there exists a
common target family {𝑋 𝛼 ⊆ 𝑈} 𝛼∈ 𝐴 such that for every 𝛼 ∈ 𝐴 and every 𝑒 ∈ 𝐿 𝛼 ∩ 𝑀 𝛼 ,
𝑒
F (2) (𝛼) = (𝐹𝛼 , 𝐿 𝛼 ), G (2) (𝛼) = (𝐺 𝛼 , 𝑀 𝛼 ), 𝐹𝛼 (𝑒) = 𝐺 𝛼 (𝑒) = 𝑋 𝛼 𝑒 , 𝑋 𝛼 .
Define T ∨ S and T ∧ S by setting, for each 𝛼 ∈ 𝐴,
𝐿 ∨𝛼 := 𝐿 𝛼 ∪ 𝑀 𝛼 , 𝐿 ∧𝛼 := 𝐿 𝛼 ∩ 𝑀 𝛼 (assumed nonempty),
and
(T ∨ S) ↓ 𝛼 := 𝐻 ∨𝛼 , 𝐿 ∨𝛼 , (T ∧ S) ↓ 𝛼 := 𝐻 ∧𝛼 , 𝐿 ∧𝛼 ,
where, for 𝑒 ∈ 𝐿 ∨𝛼 and 𝑒 ∈ 𝐿 ∧𝛼 respectively,
𝑒 𝑒
𝐻 ∨𝛼 (𝑒) := 𝑋 𝛼 𝑒 , 𝑋 𝛼 , 𝐻 ∧𝛼 (𝑒) := 𝑋 𝛼 𝑒 , 𝑋 𝛼 .
Theorem 3.10 (Well-definedness of pointwise union/intersection under a common target). Under the hypothe-
ses of Definition 3.9, both T ∨ S and T ∧ S are well-defined type-2 rough sets over PA.
𝑒
Proof. Fix 𝛼 ∈ 𝐴. For each 𝑒 ∈ 𝐿 ∨𝛼 = 𝐿 𝛼 ∪ 𝑀 𝛼 , we define 𝐻 ∨𝛼 (𝑒) as the Pawlak pair (𝑋 𝛼 𝑒 , 𝑋 𝛼 ), which is
well-defined for every 𝑒 ∈ 𝐸. Hence (𝐻 ∨𝛼 , 𝐿 ∨𝛼 ) is a type-1 rough set, i.e., belongs to Σ (1) (𝑈, 𝐸, 𝜌). Therefore
T ∨ S satisfies Definition 3.3.
Similarly, provided 𝐿 ∧𝛼 = 𝐿 𝛼 ∩ 𝑀 𝛼 is nonempty, for each 𝑒 ∈ 𝐿 ∧𝛼 the same Pawlak pair definition yields that
(𝐻 ∧𝛼 , 𝐿 ∧𝛼 ) ∈ Σ (1) (𝑈, 𝐸, 𝜌). Thus T ∧ S is also a type-2 rough set. □
4
3.2 Type-n Rough Set
A Type-𝑛 Rough Set is an 𝑛-level hierarchical parameterization that recursively maps parameters to lower-level
rough sets, ending with Pawlak lower–upper approximations of subsets.
Definition 3.11 (Type-𝑛 rough set). Let PA = (𝑈, 𝐸, 𝜌) be a parameterized approximation space. Define
recursively the collections Σ (𝑛) (𝑈, 𝐸, 𝜌) of type-𝑛 rough sets as follows.
(1) Σ (1) (𝑈, 𝐸, 𝜌) is the collection of all type-1 rough sets RS (1) (𝑋, 𝐵) from Definition 3.2.
(2) For 𝑛 ≥ 2, a type-𝑛 rough set (briefly, T𝑛RS) over PA is a pair (F (𝑛) , 𝐴𝑛 ) where 𝐴𝑛 ⊆ 𝐸 is a nonempty
primary parameter set and
F (𝑛) : 𝐴𝑛 → Σ (𝑛−1) (𝑈, 𝐸, 𝜌), 𝑎 ↦−→ F (𝑛) (𝑎) = F𝑎(𝑛−1) , 𝐿 𝑎 ,
Ð
such that 𝐿 𝑎 ⊆ 𝐸 is nonempty for every 𝑎 ∈ 𝐴𝑛 . The union 𝑎∈ 𝐴𝑛 𝐿 𝑎 is called the underlying parameter
set of the type-𝑛 rough set.
Remark 3.12. Intuitively, a type-𝑛 rough set is a hierarchical parameterization: primary parameters select
type-(𝑛 − 1) rough sets, which themselves select type-(𝑛 − 2) rough sets, and so on, until one reaches type-1
rough sets, i.e., families of Pawlak lower/upper approximations.
Example 3.13 (Real-life example of a Type-𝑛 rough set). Consider a hospital triage system that must decide
whether a patient belongs to the target class
𝑋 = {patients who require ICU admission within 24 hours} ⊆ 𝑈,
where 𝑈 is the finite set of patients arriving during a shift.
Base (type-1) level. Let 𝐸 be a set of measurement/assessment protocols (parameters), for example:
𝐸 = {𝑒 vitals , 𝑒 labs , 𝑒 imaging , 𝑒 clinical }.
For each 𝑒 ∈ 𝐸, define an equivalence relation 𝜌(𝑒) on 𝑈 by
𝑝 ∼𝜌(𝑒) 𝑞 ⇐⇒ 𝑝 and 𝑞 fall into the same category under protocol 𝑒,
e.g., under 𝑒 vitals they share the same (discretized) triage bins for heart rate, blood pressure, and oxygen
saturation. Then each 𝑒 yields a Pawlak rough approximation pair
𝑒
𝑋 𝑒, 𝑋 ,
𝑒
where 𝑋 𝑒 are patients definitely ICU-bound under protocol 𝑒, and 𝑋 are those possibly ICU-bound.
Type-2 level (departmental viewpoints). Let the primary parameter set be the set of clinical departments
𝐴2 = {ER, ICU, Cardiology}.
For each department 𝑎 ∈ 𝐴2 , let 𝐿 𝑎 ⊆ 𝐸 be the subset of protocols that department trusts most (e.g., ER
emphasizes vitals and clinical examination, ICU emphasizes labs plus vitals). Define F (2) (𝑎) to be the type-1
rough set over 𝐿 𝑎 , i.e., the family
𝑒
𝑒 ∈ 𝐿 𝑎 ↦−→ 𝑋 𝑒 , 𝑋 .
Thus, a type-2 rough set captures that different departments use different granulations and thus produce different
“definitely”/“possibly” ICU sets.
Type-𝑛 level (policy → department → protocol → · · · ). For 𝑛 ≥ 3, add higher layers of decision context, for
example:
𝐴3 = {day shift policy, night shift policy}, 𝐴4 = {normal capacity, surge capacity},
where each policy or capacity state selects (via F (3) or F (4) ) an appropriate lower-level rough set that reflects
which departments and protocols are prioritized. In this way, a type-𝑛 rough set models hierarchical uncertainty
management: high-level operational conditions choose mid-level medical viewpoints, which choose low-level
measurement protocols, eventually yielding Pawlak lower/upper approximations of 𝑋 for concrete patient
decisions.
5
Definition 3.14 (Admissible parameter chains and evaluation). Let 𝑛 ≥ 1 and let S = (F (𝑛) , 𝐴𝑛 ) ∈
Σ (𝑛) (𝑈, 𝐸, 𝜌).
(i) The set Π𝑛 (S) of admissible parameter chains is defined recursively by:
n o
Π1 (S) := 𝐴1 , Π𝑛 (S) := (𝑎, 𝜋) 𝑎 ∈ 𝐴𝑛 , 𝜋 ∈ Π𝑛−1 S↓ 𝑎 (𝑛 ≥ 2),
where S↓ 𝑎 := F (𝑛) (𝑎) ∈ Σ (𝑛−1) (𝑈, 𝐸, 𝜌) is the section at 𝑎 (well-defined by Definition 3.11).
(ii) The evaluation map evS : Π𝑛 (S) → P (𝑈) × P (𝑈) is defined recursively by:
evS (𝑒) := F (1) (𝑒) ∈ P (𝑈) × P (𝑈) (𝑛 = 1),
and for 𝑛 ≥ 2,
evS (𝑎, 𝜋) := evS↓𝑎 (𝜋).
Lemma 3.15 (Partition of chains). Let 𝑛 ≥ 2 and S = (F (𝑛) , 𝐴𝑛 ) ∈ Σ (𝑛) (𝑈, 𝐸, 𝜌). Then Π𝑛 (S) is the disjoint
union of the subsets
Π𝑛 (S; 𝑎) := {(𝑎, 𝜋) ∈ Π𝑛 (S) | 𝜋 ∈ Π𝑛−1 (S ↓ 𝑎)}, 𝑎 ∈ 𝐴𝑛 ,
and consequently ∑︁
|Π𝑛 (S)| = |Π𝑛−1 (S ↓ 𝑎)|.
𝑎∈ 𝐴𝑛
Proof. By Definition 3.14(i), every chain in Π𝑛 (S) has a unique first coordinate 𝑎 ∈ 𝐴𝑛 , hence belongs to exactly
one Π𝑛 (S; 𝑎), proving disjointness and the union claim. The cardinality identity follows immediately. □
Theorem 3.16 (Leaf roughness: every evaluation is a Pawlak approximation pair). Let 𝑛 ≥ 1 and let S ∈
Σ (𝑛) (𝑈, 𝐸, 𝜌). For every admissible chain 𝜋 ∈ Π𝑛 (S), the value evS (𝜋) = (𝐿 𝜋 , 𝑈 𝜋 ) satisfies
𝐿 𝜋 ⊆ 𝑈𝜋 .
Moreover, there exist a parameter 𝑒 𝜋 ∈ 𝐸 and a target set 𝑋 𝜋 ⊆ 𝑈 such that
𝑒𝜋
(𝐿 𝜋 , 𝑈 𝜋 ) = 𝑋 𝜋 𝑒 𝜋 , 𝑋 𝜋 ,
𝑒
where (·) 𝑒 and (·) are the Pawlak lower/upper approximations computed with respect to the equivalence
relation 𝜌(𝑒). In particular,
𝐿 𝜋 ⊆ 𝑋𝜋 ⊆ 𝑈𝜋 .
Proof. We argue by induction on 𝑛.
𝑒
Base case 𝑛 = 1. Then S = (F (1) , 𝐴1 ) is a type-1 rough set, so for each 𝑒 ∈ 𝐴1 we have F (1) (𝑒) = (𝑋 𝑒 , 𝑋 ) for
𝑒
some 𝑋 ⊆ 𝑈 (Definition 3.2). Hence evS (𝑒) is a Pawlak approximation pair, and the inclusions 𝑋 𝑒 ⊆ 𝑋 ⊆ 𝑋
𝑒
imply 𝑋 𝑒 ⊆ 𝑋 .
Inductive step. Assume the statement holds for (𝑛−1). Let 𝑛 ≥ 2 and take any 𝜋 ∈ Π𝑛 (S). By Definition 3.14(i),
we can write 𝜋 = (𝑎, 𝜋 ′ ) with 𝑎 ∈ 𝐴𝑛 and 𝜋 ′ ∈ Π𝑛−1 (S ↓ 𝑎). By Definition 3.14(ii),
evS (𝜋) = evS↓𝑎 (𝜋 ′ ).
Since S ↓ 𝑎 ∈ Σ (𝑛−1) (𝑈, 𝐸, 𝜌), the induction hypothesis applies to S ↓ 𝑎 and 𝜋 ′ , yielding the desired
representation as a Pawlak approximation pair and the inclusions. □
Theorem 3.17 (Sections are type-(𝑛 − 1) rough sets). Let 𝑛 ≥ 2 and S = (F (𝑛) , 𝐴𝑛 ) ∈ Σ (𝑛) (𝑈, 𝐸, 𝜌). For
every 𝑎 ∈ 𝐴𝑛 , the section S ↓ 𝑎 := F (𝑛) (𝑎) belongs to Σ (𝑛−1) (𝑈, 𝐸, 𝜌). More generally, for any admissible
prefix (𝑎 𝑛 , 𝑎 𝑛−1 , . . . , 𝑎 𝑘+1 ) of length (𝑛 − 𝑘), the iterated section
S ↓ 𝑎 𝑛 ↓ 𝑎 𝑛−1 ↓ · · · ↓ 𝑎 𝑘+1
is a type-𝑘 rough set in Σ (𝑘 ) (𝑈, 𝐸, 𝜌).
6
Proof. The first claim is immediate from Definition 3.11, which requires F (𝑛) (𝑎) ∈ Σ (𝑛−1) (𝑈, 𝐸, 𝜌) for each
𝑎 ∈ 𝐴𝑛 . The general statement follows by iterating the same argument along the prefix and applying induction
on the number of iterations. □
Theorem 3.18 (Closure under restriction of primary parameters). Let 𝑛 ≥ 1 and S = (F (𝑛) , 𝐴𝑛 ) ∈
Σ (𝑛) (𝑈, 𝐸, 𝜌). For any nonempty subset 𝐴′𝑛 ⊆ 𝐴𝑛 , define the restriction
S 𝐴′ := (F (𝑛) ↾ 𝐴′𝑛 , 𝐴′𝑛 ).
𝑛
Then S 𝐴′ ∈ Σ (𝑛) (𝑈, 𝐸, 𝜌). Furthermore, for any 𝑎 ∈ 𝐴′𝑛 , the section of the restricted object satisfies
𝑛
S 𝐴′ ↓ 𝑎 = S ↓ 𝑎.
𝑛
Proof. If 𝑛 = 1, then S is a type-1 rough set over 𝐴1 , and restricting the domain of F (1) to a nonempty subset
𝐴1′ clearly yields another type-1 rough set. Let 𝑛 ≥ 2. By Definition 3.11, F (𝑛) maps each 𝑎 ∈ 𝐴𝑛 to a
type-(𝑛 − 1) rough set. Hence its restriction F (𝑛) ↾ 𝐴′𝑛 maps each 𝑎 ∈ 𝐴′𝑛 to the same type-(𝑛 − 1) rough set,
and therefore (F (𝑛) ↾ 𝐴′𝑛 , 𝐴′𝑛 ) satisfies the axioms of a type-𝑛 rough set. The identity of sections is immediate
from the definition of restriction. □
Theorem 3.19 (Canonical rooted-tree representation). Let 𝑛 ≥ 1 and S ∈ Σ (𝑛) (𝑈, 𝐸, 𝜌). There is a canonical
finite rooted tree 𝑇 (S) of depth 𝑛 whose root-to-leaf paths are exactly the admissible chains in Π𝑛 (S), and
whose leaf-label at a path 𝜋 is the pair evS (𝜋) ∈ P (𝑈) × P (𝑈).
Conversely, any finite rooted tree of depth 𝑛 whose internal nodes are labeled by parameters from 𝐸 (and
whose outgoing edges enumerate the next-level parameter choices), and whose leaves are labeled by Pawlak
approximation pairs, determines (canonically) a type-𝑛 rough set in Σ (𝑛) (𝑈, 𝐸, 𝜌).
Proof. (Forward construction.) Let the vertices of 𝑇 (S) be all admissible prefixes of chains in Π𝑛 (S), including
the empty prefix as the root. Connect a prefix 𝑝 to 𝑝 ′ by an edge if 𝑝 ′ is obtained from 𝑝 by appending one
admissible next parameter. By construction, root-to-leaf paths correspond bijectively to full admissible chains
𝜋 ∈ Π𝑛 (S), and we label each leaf by evS (𝜋). Finiteness follows from finiteness of all parameter sets.
(Reverse reconstruction.) Proceed by induction on 𝑛. For 𝑛 = 1, a depth-1 tree is just a nonempty set of
parameters labeled by Pawlak pairs, which is exactly a type-1 rough set. Assume the claim holds for (𝑛 − 1).
Given a depth-𝑛 labeled tree, take the root-level parameter set 𝐴𝑛 to be the set of labels on edges leaving the
root, and for each 𝑎 ∈ 𝐴𝑛 take the rooted subtree 𝑇𝑎 hanging from that edge. By the induction hypothesis, each
𝑇𝑎 determines a type-(𝑛 − 1) rough set, and setting F (𝑛) (𝑎) equal to that type-(𝑛 − 1) object yields a type-𝑛
rough set. Canonicity follows from the deterministic nature of the construction. □
4 Conclusion
In this paper, inspired by the hierarchy of Type-2 and, more generally, Type-𝑛 fuzzy sets, we investigated
corresponding generalizations of rough sets, namely Type-2 and Type-𝑛 rough sets. In future work, we expect
further developments that extend these concepts by incorporating uncertainty-aware frameworks such as fuzzy
sets [1], neutrosophic sets [3, 21], and plithogenic sets [5].
Funding
This study did not receive any financial or external support from organizations or individuals.
Acknowledgments
We extend our sincere gratitude to everyone who provided insights, inspiration, and assistance throughout this
research. We particularly thank our readers for their interest and acknowledge the authors of the cited works
for laying the foundation that made our study possible. We also appreciate the support from individuals and
institutions that provided the resources and infrastructure needed to produce and share this paper. Finally, we
are grateful to all those who supported us in various ways during this project.
7
Data Availability
This research is purely theoretical, involving no data collection or analysis. We encourage future researchers
to pursue empirical investigations to further develop and validate the concepts introduced here.
Ethical Approval
As this research is entirely theoretical in nature and does not involve human participants or animal subjects, no
ethical approval is required.
Conflicts of Interest
The authors confirm that there are no conflicts of interest related to the research or its publication.
Disclaimer
This work presents theoretical concepts that have not yet undergone practical testing or validation. Future
researchers are encouraged to apply and assess these ideas in empirical contexts. While every effort has been
made to ensure accuracy and appropriate referencing, unintentional errors or omissions may still exist. Readers
are advised to verify referenced materials on their own. The views and conclusions expressed here are the
authors’ own and do not necessarily reflect those of their affiliated organizations.
References
[1] Lotfi A Zadeh. Fuzzy sets. Information and control, 8(3):338–353, 1965.
[2] Krassimir T Atanassov. Circular intuitionistic fuzzy sets. Journal of Intelligent & Fuzzy Systems, 39(5):5981–5986, 2020.
[3] Haibin Wang, Florentin Smarandache, Yanqing Zhang, and Rajshekhar Sunderraman. Single valued neutrosophic sets. Infinite study,
2010.
[4] Pradip Kumar Maji, Ranjit Biswas, and A Ranjan Roy. Soft set theory. Computers & mathematics with applications, 45(4-5):555–562,
2003.
[5] Florentin Smarandache. Plithogenic set, an extension of crisp, fuzzy, intuitionistic fuzzy, and neutrosophic sets-revisited. Infinite
study, 2018.
[6] Zdzis law Pawlak. Rough sets. International journal of computer & information sciences, 11:341–356, 1982.
[7] Takaaki Fujita. Short introduction to rough, hyperrough, superhyperrough, treerough, and multirough set. Advancing Uncertain
Combinatorics through Graphization, Hyperization, and Uncertainization: Fuzzy, Neutrosophic, Soft, Rough, and Beyond, page
394, 2025.
[8] Nayani Sateesh, Pasupureddy Srinivasa Rao, and Davuluri Rajya Lakshmi. Optimized ensemble learning-based student’s performance
prediction with weighted rough set theory enabled feature mining. Concurrency and Computation: Practice and Experience,
35(7):e7601, 2023.
[9] Nagaraju Aitha and Ramachandram Srinadas. A strategy to reduce the control packet load of aodv using weighted rough set model
for manet. The International Arab Journal of Information Technology, 8(1):108–117, 2009.
[10] Chih-Ching Hsiao, Chen-Chia Chuang, Jin-Tsong Jeng, and Shun-Feng Su. A weighted fuzzy rough sets based approach for rule
extraction. In The SICE Annual Conference 2013, pages 104–109. IEEE, 2013.
[11] Nouman Azam and JingTao Yao. Analyzing uncertainties of probabilistic rough set regions with game-theoretic rough sets.
International journal of approximate reasoning, 55(1):142–155, 2014.
[12] Yiyu Yao. Two semantic issues in a probabilistic rough set model. Fundamenta Informaticae, 108(3-4):249–265, 2011.
[13] Salvatore Greco, Benedetto Matarazzo, Roman Slowinski, and Jerzy Stefanowski. Variable consistency model of dominance-based
rough sets approach. In Rough Sets and Current Trends in Computing: Second International Conference, RSCTC 2000 Banff,
Canada, October 16–19, 2000 Revised Papers 2, pages 170–181. Springer, 2001.
[14] Nilesh N. Karnik and Jerry M. Mendel. Centroid of a type-2 fuzzy set. Inf. Sci., 132:195–220, 2001.
[15] John T. Rickard, Janet Aisbett, and Ty Rickard. On a class of general type-n normal fuzzy sets synthesized from subject matter
expert inputs. IEEE Transactions on Fuzzy Systems, 32:3178–3188, 2024.
[16] Quanguo Bi, Yu Chen, Weizhong Wang, Muhammet Deveci, and Dursun Delen. Assessing middle managers’ digital competency in
manufacturing transformation: An integrated type-2 neutrosophic exptodim approach. Applied Soft Computing, page 114276, 2025.
[17] Hend Ahmed. Generative ai in soft computing: Methodological implications and a literature survey on mcdm and type-2 neutrosophic
applications. International Journal of Computers and Informatics (Zagazig University), 9:17–26, 2025.
[18] Khizar Hayat, Muhammad Irfan Ali, Bing-Yuan Cao, and Faruk Karaaslan. New results on type-2 soft sets. Hacettepe Journal of
Mathematics and Statistics, 47(4):855–876, 2018.
[19] Zdzislaw Pawlak. Rough set theory and its applications to data analysis. Cybernetics & Systems, 29(7):661–688, 1998.
[20] Zdzis law Pawlak. Rough sets: Theoretical aspects of reasoning about data, volume 9. Springer Science & Business Media, 2012.
[21] Said Broumi, Mohamed Talea, Assia Bakali, and Florentin Smarandache. Interval valued neutrosophic graphs. Critical Review, XII,
2016:5–33, 2016.