Original Scientific Article Journal of Decision Analytics and Intelligent Computing
Vol. 5 issue 1, (2025) 219-228 [Link]
Fuzzy score and fuzzy cost SuperHyperFunction
Takaaki Fujita1,* and Arif Mehmood2
1 Independent Researcher (not affiliated with any university or research institute), Tokyo, Japan
2 Department of Mathematics, Institute of Numerical Sciences, Gomal University, Pakistan
* Correspondence: Takaaki.fujita060@[Link]
Received 17 July 2025
Accepted for publication 22 November 2025
Published 30 November 2025
Abstract
A HyperFunction associates each input with a set of admissible outputs, extending conventional functions by
allowing multi-valued rather than single-valued mappings. A SuperHyperFunction further generalizes this concept
by employing iterated powersets for both its domain and codomain, enabling the representation of hierarchical,
multi-level output structures and HyperStructural multi-valued behavior within complex systems. Although
HyperFunctions and SuperHyperFunctions offer expressive tools for modeling hierarchical functional
relationships, their study in the existing literature remains relatively limited. This paper extends the fuzzy score
function and fuzzy cost function within the frameworks of HyperFunctions and SuperHyperFunctions and provides
a concise theoretical analysis of their essential properties.
Keywords: fuzzy function, fuzzy score function, fuzzy cost function.
1. Introduction
A classical function selects exactly one value in the codomain for each element of its domain, thereby encoding
a deterministic input–output rule (Bylinski, 1990). A hyperfunction generalizes this notion by assigning to each
input a set of possible outputs, i.e., a multi-valued image (Fujita et al., 2025). An n-SuperHyperFunction goes
further: it maps subsets drawn from iterated powersets to higher-level powersets, supporting hierarchical, multi-
level outputs and HyperStructural behavior across domains (Smarandache, 2022). These ideas have been
examined in recent studies (Jdid et al., 2025; Zheng and Wang, 2025; Chen et al., 2025; Huang et al., 2025; Al-
Odhari, 2025).
Various mathematical frameworks have been developed to handle uncertainty, including the Fuzzy Set (Zadeh,
1965), Intuitionistic Fuzzy Set (Atanassov, 2012), Single-valued Neutrosophic Set (Wang et al., 2010; Xu et al., 2025;
Wang and Yang, 2025), Multipolar Neutrosophic set (Naveed and Ali, 2024; Du, 2025; Wu, 2025), Complex
neutrosophic set (Ali and Smarandache, 2017; Rajalakshmi et al., 2025; Saini et al., 2025), and Plithogenic Set
(Smarandache, 2018; Fujita, 2025d; Amable et al., 2025; Junco, 2025). A fuzzy function maps inputs either to
membership grades in [0,1] or to fuzzy sets, preserving uncertainty and enabling approximate reasoning (Demirci,
1999; Perfilieva, 2004). Well-known variants include the Intuitionistic Fuzzy Functions (Tak, 2020), Neutrosophic
Functions (Hatip, 2020; Vadivel and Sundar, 2021; Basker and Said, 2023), and Plithogenic Functions (Alhasan and
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
Abdulfatah, 2023). A fuzzy score function aggregates memberships, weights, and criteria into a single scalar to
rank alternatives under vagueness and partial truth. A fuzzy cost function represents uncertain costs by fuzzy
numbers, supporting decisions that minimize expected expense under imprecise, context-dependent information.
Although HyperFunctions and SuperHyperFunction provide expressive tools for modeling hierarchical
functional behavior, the literature remains comparatively sparse. In this paper, we extend the fuzzy score function
and fuzzy cost function within the HyperFunction and SuperHyperFunction frameworks and offer a brief analysis
of their basic properties.
2. Preliminaries
This section presents the key concepts and definitions required for the discussions in this paper. Unless
otherwise stated, all sets and structures considered here are assumed to be finite and simple (undirected, no
loops). Note that, unless otherwise specified, n denotes a natural number.
2.1 HyperFunction and n-SuperHyperFunction
Within the study of HyperStructures (Brown et al., 1992; Davvaz and Vougiouklis, 2018; Davvaz, 2020;
Agusfrianto et al., 2024) and n-SuperHyperStructures (Smarandache, 2024; Fujita, 2025b) for functions, the
notions of HyperFunction and n-SuperHyperFunction were formulated by Smarandache (2022). Since then,
HyperFunctions have attracted substantial attention and a variety of applications have been explored. For
completeness, the essential definitions and related theorems are summarized below.
Definition 2.1 (Base Set). A (finite) base set X is the foundational set from which complex structures such as
powersets and HyperStructures are derived. It is formally defined as:
X = {x | x is an element within a specified domain} (1)
Definition 2.2 (Powerset) (Fujita, 2025a). The (finite) powerset of a set X, denoted P(X), is the collection of all
possible subsets of X, including both the empty set and X itself. Formally, it is expressed as:
P(X) = {A | A ⊆ X}. (2)
Definition 2.3 (n-th Powerset) (Smarandache, 2017; Smarandache, 2022) The (finite) n-th powerset of a set X,
denoted Pn(X), is defined iteratively, starting with the standard powerset. The recursive construction is given by:
P1(X) = P(X); Pn+1(X) = P(Pn(X)), for n≥ 1 (3)
*
Similarly, the(finite) n-th non-empty powerset, denoted P n(X), is defined recursively as:
P*1(X) = P(X) \ {∅}; P*n+1) (X) = P(P*n(X)) \ {∅} (4)
Example 2.4: Let X = {Alice, Bob, Carol} be the set of three employees in an IT company.
1. P1(X) = P(X) is the set of all possible teams that can be formed from these employees: P1(X) = {∅, {Alice},
{Bob}, {Carol}, {Alice, Bob}, {Alice, Carol}, {Bob, Carol}, {Alice, Bob, Carol}}.
2. P2(X) = P(P1(X)) is the set of all possible “collections of teams”. For example, {{Alice}, {Bob, Carol}} ∈ P2(X)
means: “one task is done by Alice alone, and another task is done by the team {Bob, Carol}.” Thus P2(X)
can model one project made of several subteams.
In practice, P1(X) corresponds to “all possible teams from the company,” while P2(X) corresponds to “all
possible ways to group those teams into a multi-team project.” Higher powersets P3(X), P4(X), … can then
represent portfolios of projects, or layers of management that group projects into larger programs.
Definition 2.5 (HyperOperation) (Spartalis, 1996; Vougiouklis, 2005; Leoreanu-Fotea et al., 2015): A (finite)
HyperOperation is a generalization of a binary operation where the result of combining two elements is a set, not
a single element. Formally, for a set X, a HyperOperation ◦ is defined as:
◦: X × X → P(X) (5)
220
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
Definition 2.6 (HyperFunction) (Fujita et al., 2025; Fujita, 2025c). A (finite) HyperFunction is a function where
the domain remains a classical set X, but the codomain is extended to the powerset of X, denoted P(X). Formally,
a HyperFunction f is defined as:
f: X → P(X). For any x ∈ X, f(x) ⊆ X is a subset of X (6)
Example 2.7 : Let X = {email, chat, phone} be the possible contact channels for a customer. Define
f: X → P(X) by
f(email) = {email, chat}, f(chat) = {chat}, f(phone) = {phone, email}.
This models a helpdesk rule: if a customer writes by email, the system may respond by email and additionally
open a chat; if the customer calls by phone, the system may register both a phone response and a confirmation
email. One input (a single channel) is mapped to several possible output channels (a set), so f is a HyperFunction.
Definition 2.8 (SuperHyperOperations) (Fujita, 2025b). Let X be a (finite) non-empty set, and let Pk(X) be the
(finite) k-th powerset of X. Define
𝑃0(𝐻) = 𝐻, 𝑃𝑘+1(𝐻) = 𝑃 (𝑃𝑘 (𝐻)) for 𝑘 ≥ 0 (7)
A (finite) SuperHyperOperation of order (m, n) is an m-ary operation
◦ (m, n): Hm → P*n(X) (8)
If the codomain excludes the empty set, it is classical-type; if it includes it, it is Neutrosophic-type.
Definition 2.9 (n-SuperHyperFunction) (Smarandache, 2022). An (finite) n-SuperHyperFunction generalizes the
concept of a HyperFunction by using the n-th powerset Pn(S) as the codomain. Formally, for n ≥ 2, an n-
SuperHyperFunction f is defined as:
f: Pr(S) → Pn(S) (9)
where 0 ≤ r ≤ n, and Pn(S) is the n-th powerset of S. This definition allows f to map subsets of S (from Pr(S)) to
elements in the n-th powerset Pn(S).
Example 2.10.: Let S = {HR, Accounting, IT} be the departments of a company. A request may involve several
departments at once (for example {HR, IT}), and each such joint request must be expanded into several “bundles”
of subworkflows (for example a bundle for approval, a bundle for security checks, and a bundle for payroll update).
Define
g: P(S) → P₂(S)
that sends a subset like {HR, IT} to a collection of subsets of P(S), for example
g({HR, IT}) = {{{HR}, {IT}}, {{HR, Accounting}}}.
This means: the joint request {HR, IT} will be realized as (i) one workflow where HR and IT act separately, and
(ii) an alternative workflow where HR must cooperate with accounting. Since the codomain is the second powerset
P₂(S), g is an n-SuperHyperFunction with n = 2.
2.2. Fuzzy score function
A score function maps an object to a single scalar, enabling total ordering and comparison across alternatives
under defined criteria (Nancy and Garg, 2016). A fuzzy score function assigns each fuzzy quantity a scalar,
aggregating membership information to rank uncertain alternatives consistently and comparably (Kumar et al.,
2025).
Definition 2.11 (Lee, 2024): A fuzzy score function is a mapping S: F → R that assigns each fuzzy quantity
(e.g., a fuzzy number or an element of an extended fuzzy set model) a single scalar so that alternatives can be
totally or quasi-totally ordered under uncertainty. Good score functions are designed to increase with “degree of
membership” and, in models that include non-membership, to decrease with “degree of non-membership,” often
normalizing to [0,1] for comparability.
Example 2.12: Suppose two suppliers A and B are evaluated by triangular fuzzy numbers (TFNs):
QA = TFN(0.6, 0.8, 0.9)
221
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
QB = TFN(0.5, 0.7, 0.95)
Define the fuzzy score function S for a triangular fuzzy number TFN(a, b, c) by
S(TFN(a, b, c)) = (a + b + c) / 3.
Then
S(QA) = (0.6 + 0.8 + 0.9) / 3 = 2.3 / 3 ≈ 0.7667,
S(QB) = (0.5 + 0.7 + 0.95) / 3 = 2.15 / 3 ≈ 0.7167.
Since 0.7667 > 0.7167, supplier A ranks above supplier B under uncertainty.
2.3. Fuzzy cost function
A cost function assigns each decision or model a numeric penalty, measuring error or expense to guide
optimization and comparison. A fuzzy cost function assigns fuzzy costs to alternatives, aggregating uncertain
criteria and enabling ranking through defuzzification or fuzzy ordering (Garcia et al., 2009).
Definition 2.13 (Vasant et al., 2006; Jain and Sanga, 2020): A fuzzy cost function maps each decision, path, or
design xxx to a fuzzy valuation C(x)∈R~ (or to a fuzzy rule–based score), aggregating uncertain criteria such as
distance, risk, or energy through fuzzy membership functions and inference. Decisions are then compared by
defuzzifying or ranking these fuzzy costs. In practice, such functions have been used to evaluate and choose robot
paths by fusing multiple uncertain factors into a single fuzzy “cost” for optimization.
Example 2.14: A mobile robot must choose between two paths P1 and P2. Each path has an uncertain
(normalized) distance and risk modeled by triangular fuzzy numbers. Distances (in km) are normalized by the scale
12 km.
Path P1: distance D1 = TFN(8, 10, 12) / 12 = TFN(2/3, 5/6, 1) ≈ TFN(0.6667, 0.8333, 1.0000), risk R1 = TFN(0.2, 0.3,
0.5).
Path P2:distance D2 = TFN(6, 9, 11) / 12 = TFN(0.5, 0.75, 11/12) ≈ TFN(0.5000, 0.7500, 0.9167), risk R2 = TFN(0.3,
0.4, 0.6).
Define a fuzzy cost function with weights wd = 0.7 (distance) and wr = 0.3 (risk) using TFN arithmetic:
Scalar multiply: α · TFN(a, b, c) = TFN(αa, αb, αc).
Componentwise add: TFN(a1, b1, c1) ⊕ TFN(a2, b2, c2) = TFN(a1 + a2, b1 + b2, c1 + c2).
Set C(P) = wd · D ⊕ wr · R.
Compute: C(P1) = 0.7 · TFN(0.6667, 0.8333, 1.0000) ⊕ 0.3 · TFN(0.2, 0.3, 0.5)
= TFN(0.4667, 0.5833, 0.7000) ⊕ TFN(0.0600, 0.0900, 0.1500)
= TFN(0.5267, 0.6733, 0.8500).
C(P2) = 0.7 · TFN(0.5000, 0.7500, 0.9167) ⊕ 0.3 · TFN(0.3, 0.4, 0.6)
= TFN(0.3500, 0.5250, 0.6417) ⊕ TFN(0.0900, 0.1200, 0.1800)
= TFN(0.4400, 0.6450, 0.8217).
Defuzzify by the centroid: cen(TFN(a, b, c)) = (a + b + c) / 3.
cen(C(P1)) = (0.5267 + 0.6733 + 0.8500) / 3 ≈ 0.6833,
cen(C(P2)) = (0.4400 + 0.6450 + 0.8217) / 3 ≈ 0.6356.
Because 0.6356 < 0.6833, the fuzzy cost function selects P2 as the better (lower-cost) path under uncertainty.
3. Main Results
This section presents the main contributions of the paper.
3.1. Fuzzy Score HyperFunction
A Fuzzy Score HyperFunction assigns each fuzzy object the set of all admissible scalar scores, capturing modeling
or parameter uncertainty.
222
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
Definition 3.1.: Fix a nonempty family S of admissible fuzzy score maps on F, i.e.,
S ⊆ {s: F → R} with S ≠ ∅ (10)
where each s ∈ S satisfies the modeling requirements for a score (for example, it is nondecreasing in “membership
degree,” nonincreasing in “non-membership,” and optionally normalized to [0,1]).
Define the Fuzzy Score HyperFunction associated with S by
HS: F → P*(R) (11)
given by
HS(x) := {s(x): s ∈ S} for every x ∈ F (12)
Intuition: HS(x) collects all admissible scalar scores of x produced by the family S (e.g., different defuzzification
rules, parameter settings, or weighting schemes). Thus, HS is multi-valued in the codomain and captures score
uncertainty or modeling ambiguity.
Example 3.2.: Let F be the set of triangular fuzzy numbers TFN(a, b, c) with a ≤ b ≤ c in [0,1]. For λ ∈ [0,1], define
an admissible score: sλ(TFN(a, b, c)):= (a + λb + c) / (2 + λ). Interpretation: λ interpolates between the simple
average of the endpoints (λ = 0) and the usual centroid (λ = 1). Set S := { sλ : λ ∈ [0,1] } and define HS accordingly.
Compare two suppliers with quality TFNs:
QA = TFN(0.6, 0.8, 0.9)
QB = TFN(0.5, 0.7, 0.95)
For QA:
s0(QA) = (0.6 + 0.9) / 2 = 0.75
s1(QA) = (0.6 + 0.8 + 0.9) / 3 = 2.3 / 3 ≈ 0.7667
Hence HS(QA) ⊇ [0.75, 0.7667] (indeed, the full image over λ ∈ [0,1] is a compact interval between these
endpoints).
For QB:
s0(Q_B) = (0.5 + 0.95) / 2 = 0.725
s1(Q_B) = (0.5 + 0.7 + 0.95) / 3 = 2.15 / 3 ≈ 0.7167
Hence H_S(Q_B) ⊇ [0.7167, 0.725] (again an interval).
Dominance: note that min HS(QA) = 0.75 > 0.725 = max HS(QB). Therefore, QA robustly outranks QB for all λ ∈
[0,1]. This illustrates how the HyperFunction returns a set of plausible scores, while still enabling clear decisions
when the sets are separated.
Theorem 3.3.: For any nonempty family S ⊆ {s: F → R}, the map HS: F → P*(R) defined by HS(x) = {s(x): s ∈ S} is a
HyperFunction.
Proof. Fix x ∈ F. Since S ≠ ∅ and every s ∈ S maps x to a real number s(x) ∈ R, the image set HS(x) = {s(x): s ∈ S} is
a nonempty subset of R. Hence HS(x) ∈ P*(R). Because this holds for every x ∈ F, HS is a function from F to P*(R),
i.e., a HyperFunction.
Theorem 3.4.: (i) For every (crisp) fuzzy score function s: F → R, there exists a Fuzzy Score HyperFunction HS
whose values are singletons and which “embeds” s. (ii) Conversely, any single-valued selection from a Fuzzy Score
HyperFunction yields a (crisp) fuzzy score function.
Proof. (i) Take S = {s}. Then for each x ∈ F, HS(x) = {s(x)}, which is a singleton subset of R. Thus s can be recovered
by the canonical “projection” s(x) = the unique element of HS(x). Hence every ordinary fuzzy score function is a
special case of the Fuzzy Score HyperFunction.
Proof. (ii) Let S be any nonempty family and HS the associated HyperFunction. Suppose we choose, for each x ∈
F, one element σ(x) ∈ HS(x) (i.e., a single-valued choice). Then the map sσ: F → R defined by sσ(x) = σ(x) is an ordinary
fuzzy score function (it assigns exactly one scalar to each x). Therefore, single-valued selections of HS recover (crisp)
fuzzy score functions.
223
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
3.2. Fuzzy Score SuperHyperFunction
A Fuzzy Score SuperHyperFunction maps fuzzy subsets to higher-level nested score families, aggregating multi-
stage evaluations and preserving non-emptiness across levels.
Notation 3.5. (Singleton nesting (level raising)): For any set Z and integer t ≥ 0 define Nestt(Z) by:
Nest0(Z):= Z
Nestt+1(Z):= {Nestt(Z)}
We will use the following simple facts (by construction):
If Z ∈ Pj(Y) and t ≥ 0, then Nestt(Z) ∈ Pj+t(Y).
If S ≠ ∅, then {something depending on s ∈ S} is nonempty.
Definition 3.6.: Let S be a nonempty family of admissible fuzzy score maps on F, i.e.,
S ⊆ {s: F → R} with S ≠ ∅ (13)
Fix integers m,n with n ≥ m+1 (so at least one level of “hyper” aggregation appears).
Define the Fuzzy Score SuperHyperFunction associated with S by
HS(m,n): Pm(F) → Pn(R) (14)
For X ∈ Pm(F), set
CS(m)(X):= {Pm(s)(X) : s ∈ S} ∈ Pm+1(R) (15)
and then
HS(m,n)(X):= Nestn-(m+1)(CS(m)(X) ) ∈ Pn(R) (16)
Intuition:
• Pm(s) applies “score s” pointwise at level m (via direct image), producing an element of Pm(R).
• Collecting over all s ∈ S yields CS(m)(X) ∈ Pm+1(R).
• Additional Nest raises the level to n, encoding the super-hyper structure.
Because S ≠ ∅, each HS(m,n)(X) is nonempty.
Example 3.7. (Level lift with m = 1 and n = 2): Let F be the set of triangular fuzzy numbers TFN(a,b,c) with 0 ≤ a
≤ b ≤ c ≤ 1. For λ ∈ [0,1], define an admissible score sλ(TFN(a,b,c)) := (a + λ b + c) / (2 + λ). Let S := {sλ : λ ∈ {0, 1}}
(two scoring rules: endpoints-average and centroid).
Take a finite set X ∈ P1(F) of candidates, e.g., X = {TFN(0.6,0.8,0.9), TFN(0.5,0.7,0.95)}. Compute P1(sλ)(X) = { sλ(x)
: x ∈ X } ⊆ R, so P1(sλ)(X) ∈ P1(R). Thus CS(1)(X) = {P1(s0)(X), P1(s1)(X)} ∈ P2(R), and for n = 2 we have HS(1,2)(X) = Nest2-
(1+1)
(CS(1)(X)) = CS(1)(X) ∈ P2(R).
So HS(1,2) returns a set of (score-sets), one per scoring rule, i.e., a 2-level super-hyper score.
Theorem 3.8.: For any nonempty S ⊆ {s: F → R} and integers m,n with n ≥ m+1, the map HS(m,n) : Pm(F) → Pn(R)
defined above is a (m,n)–SuperHyperFunction.
Proof. Fix X ∈ Pm(F). For each s ∈ S, Pm(s)(X) ∈ Pm(R) by the definition of the direct image lift. Therefore CS(m)(X)
:= {Pm(s)(X) : s ∈ S} ⊆ Pm(R), hence CS(m)(X) ∈ Pm+1(R). Since n ≥ m+1, applying Nestn-(m+1) maps Pm+1(R) into Pn(R).
Thus HS(m,n)(X) ∈ Pn(R). As S ≠ ∅, CS(m)(X) is nonempty, and so is HS(m,n)(X). Therefore HS(m,n) is a well-typed function
Pm(F) → Pn(R), i.e., a (m,n)–SuperHyperFunction.
Theorem 3.9.: The Fuzzy Score SuperHyperFunction strictly generalizes the Fuzzy Score HyperFunction. In
particular, HS(0,1) = HS.
Proof. Set m = 0 and n = 1. Then P0(F) = F and, for X ∈ F, P0(s)(X) = s(X) ∈ R. Hence CS(0)(X) = {P0(s)(X): s ∈ S} =
{s(X): s ∈ S} ⊆ R, so CS(0)(X) ∈ P1(R) = P(R). Because n-(m+1) = 1-(0+1) = 0, we have HS(0,1)(X) = Nest0(CS(0)(X)) = CS(0)(X)
= {s(X) : s ∈ S} = HS(X). Thus the classical Fuzzy Score HyperFunction is the special case (m,n) = (0,1). Moreover,
whenever S = {s} is a singleton, HS(0,1)(x) = {s(x)} is singleton-valued, recovering the degenerate crisp score as a
subcase. Therefore the SuperHyperFunction strictly generalizes the HyperFunction.
224
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
3.3. Fuzzy Cost HyperFunction
A Fuzzy Cost HyperFunction maps each item to the nonempty set of fuzzy cost values generated by alternative
admissible models.
Definition 3.10.: Let Θ be a nonempty family of admissible fuzzy cost models on X, that is Θ ⊆ {C: X → FN(R)}
and Θ ≠ ∅. Define the Fuzzy Cost HyperFunction associated with Θ by HΘ : X → P*(FN(R)).
For each x ∈ X, set HΘ(x) := {C(x): C ∈ Θ}. (Thus HΘ(x) is the nonempty set of all fuzzy costs produced at x by the
models in Θ.)
If a defuzzifier d : FN(R) → R is fixed, also define the defuzzified Fuzzy Cost HyperFunction by HΘd : X → P*(R)
HΘd(x) := {d(C(x)): C ∈ Θ}.
Example 3.11.: Let X = {P₁, P₂} be two candidate robot paths. For each path P, define a fuzzy distance DP ∈ FN(R)
and a fuzzy risk RP ∈ FN(R). For a weight parameter w ∈ [0,1], define a fuzzy cost model C(w)(P)= w ⊗ DP ⊕ (1 −
w)⊗RP, where “⊗” is scalar multiplication of fuzzy numbers and “⊕” is componentwise addition (e.g., for
triangular/trapezoidal fuzzy numbers).
Take Θ = {C(0.7), C(0.5)}. Then the Fuzzy Cost HyperFunction is HΘ(P) = {C(0.7)(P), C(0.5)(P)} ⊆ FN(R), so HΘ(P) explicitly
captures the model uncertainty stemming from the weighting choice. If we fix the centroid defuzzifier d, the
defuzzified HyperFunction becomes HΘd(P) = {d(C(0.7)(P)), d(C(0.5)(P))} ⊆ R, representing the set of plausible crisp
costs for path P.
Theorem 3.12.: With Θ ≠ ∅ as above, HΘ: X → P*(FN(R)) is a hyperfunction.
Proof. For any fixed x ∈ X and C ∈ Θ, we have C(x) ∈ FN(R). Therefore HΘ(x) = {C(x) : C ∈ Θ} ⊆ FN(R). Because
Θ≠∅, the set HΘ(x) is nonempty, hence HΘ(x) ∈ P*(FN(R)). Thus, HΘ is a function from X into P*(FN(R)), i.e., a
HyperFunction.
Theorem 3.13.: The Fuzzy Cost HyperFunction strictly generalizes the ordinary fuzzy cost function. In particular,
if Θ is a singleton {C₀} with C₀ : X → FN(R), then H {C₀}(x) = {C₀(x)} for all x ∈ X, so H{C₀} is exactly the point-embedding
of C₀ into P*(FN(R)). Moreover, for any defuzzifier d, H{C₀}d(x) = {d(C₀(x))} embeds the crisp cost d∘C₀ into P*(R).
Proof. If Θ = {C₀}, then for each x ∈ X, H{C₀}(x) = { C(x): C ∈ {C₀} } = { C₀(x) }. Thus H{C₀} equals the map x ↦ {C₀(x)},
i.e., the canonical inclusion of C₀ into the HyperFunction codomain. Likewise, H{C₀} d(x) = {d(C(x)): C ∈ {C₀}} =
{d(C₀(x))}, which is the canonical inclusion of the crisp cost d∘C₀ into P*(R). Hence the Fuzzy Cost HyperFunction
reduces to the ordinary fuzzy cost function (or its defuzzified version) when Θ is a singleton, and therefore strictly
generalizes it when |Θ| > 1.
3.4. Fuzzy Cost SuperHyperFunction
A Fuzzy Cost SuperHyperFunction lifts cost evaluation to higher powerset levels, aggregating model-dependent
fuzzy costs into nested multi-stage structures representations.
Definition 3.14.(Fuzzy Cost HyperFunction): Let Θ be a nonempty family of admissible fuzzy cost models on X,
that is Θ ⊆ {C : X → FN(R)} and Θ ≠ ∅. Define the Fuzzy Cost HyperFunction associated with Θ by HΘ: X → P*(FN(R))
For each x ∈ X, set HΘ(x):={ C(x) : C ∈ Θ}. (Thus H_Θ(x) is the nonempty set of all fuzzy costs produced at x by the
models in Θ.)
If a defuzzifier d: FN(R) → R is fixed, also define the defuzzified Fuzzy Cost HyperFunction by HΘ : X → P*(R)
HΘd(x):= {d(C(x)) : C ∈ Θ}.
Example 3.15.: Let X = {P₁, P₂} be two candidate robot paths. For each path P, define a fuzzy distance DP ∈ FN(R)
and a fuzzy risk RP ∈ FN(R). For a weight parameter w ∈ [0,1], define a fuzzy cost model C(w)(P):= w ⊗ DP ⊕ (1 −
w) ⊗ RP, where “⊗” is scalar multiplication of fuzzy numbers and “⊕” is componentwise addition (e.g., for
triangular/trapezoidal fuzzy numbers).
225
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
Take Θ = {C(0.7), C(0.5)}. Then the Fuzzy Cost HyperFunction is HΘ(P) = {C(0.7)(P), C(0.5)(P)} ⊆ FN(R), so HΘ(P) explicitly
captures the model uncertainty stemming from the weighting choice. If we fix the centroid defuzzifier d, the
defuzzified HyperFunction becomes HΘd(P) = {d(C(0.7)(P)), d(C(0.5)(P))} ⊆ R, representing the set of plausible crisp
costs for path P.
Theorem 3.16.: With Θ ≠ ∅ as above, HΘ : X → P*(FN(R)) is a HyperFunction.
Proof. For any fixed x ∈ X and C ∈ Θ, we have C(x) ∈ FN(R). Therefore HΘ(x) = {C(x) : C ∈ Θ} ⊆ FN(R). Because
Θ≠∅, the set HΘ(x) is nonempty, hence HΘ(x) ∈ P*(FN(R)). Thus, HΘ is a function from X into P*(FN(R)), i.e., a
HyperFunction.
Theorem 3.17.: The Fuzzy Cost HyperFunction strictly generalizes the ordinary fuzzy cost function. In particular,
if Θ is a singleton {C₀} with C₀ : X → FN(R), then
H{C₀}(x) = {C₀(x)} for all x ∈ X, so H{C₀} is exactly the point-embedding of C₀ into P*(FN(R)). Moreover, for any
defuzzifier d, H{C₀}d(x) = {d(C₀(x))} embeds the crisp cost d∘C₀ into P*(R).
Proof. If Θ = {C₀}, then for each x ∈ X, H{ {C₀} }(x) = {C(x): C ∈ {C₀} } = { C₀(x)}. Thus H_{{C₀}} equals the map
x↦{C₀(x)}, i.e., the canonical inclusion of C₀ into the HyperFunction codomain. Likewise, H{C₀}d(x) = {d(C(x)) : C∈{C₀}}
= {d(C₀(x))}, which is the canonical inclusion of the crisp cost d∘C₀ into P*(R). Hence the Fuzzy Cost HyperFunction
reduces to the ordinary fuzzy cost function (or its defuzzified version) when Θ is a singleton, and therefore strictly
generalizes it when |Θ| > 1.
4. Conclusion
This paper has extended the fuzzy score function and fuzzy cost function within the frameworks of
HyperFunctions and SuperHyperFunctions and provides a concise theoretical analysis of their essential properties.
I would like to explore possible extensions in the future using Neutrosophic Functions and Plithogenic Functions.
References
Agusfrianto, F. A., Andromeda, S., & Hariri, M. (2024). Bi-Hyperstructures in Chemical Hyperstructures of Redox
Reactions with Three and Four Oxidation States. Jurnal Teori dan Aplikasi Matematika, 8(1), 50-57.
Alhasan, Y. A., & Abdulfatah, R. A. (2023). Plithogenic functions value. Neutrosophic Sets and Systems, 58, 596-
602.
Ali, M., & Smarandache, F. (2017). Complex neutrosophic set. Neural computing and applications, 28(7), 1817-
1834.
Al-Odhari, A. (2025). A brief comparative study on hyperstructure, super hyperstructure, and n-super
superhyperstructure. Neutrosophic Knowledge, 6, 38-49.
Amable, N. H., De Salazar, E. E. V., Isaac, M. G. M., Sánchez, O. C. O., & Palma, J. M. S. (2025). Representation of
motivational dynamics in school environments through Plithogenic n-SuperHyperGraphs with family
participation. Neutrosophic Sets and Systems, 92, 570-583.
Atanassov, K. T. (2012). On intuitionistic fuzzy sets theory. Part of the book series: Studies in Fuzziness and Soft
Computing, vol. 283. Springer.
Basker, P., & Said, B. (2023). On Neutrosophic Homeomorphisms via Neutrosophic Functions. Neutrosophic Sets
and Systems, 55, 403-414.
Brown, R. D., Downs, G. M., Willett, P., & Cook, A. P. (1992). Hyperstructure model for chemical structure handling:
generation and atom-by-atom searching of hyperstructures. Journal of chemical information and computer
sciences, 32(5), 522-531.
Bylinski, C. (1990). Functions and their basic properties. Formalized Mathematics, 1(1), 55-65.
Chen, Z., Qin, L., Sun, P., & Guo, W. (2025). Investment Benefit of Grid Infrastructure Projects: A Neutrosophic
SuperHyperFunction-Based Model. Neutrosophic Sets and Systems, 93, 463-473.
226
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
Davvaz, B. (2020). A brief survey on algebraic hyperstructures: Theory and applications. Journal of Algebraic
Hyperstructures and Logical Algebras, 1(3), 15-29.
Davvaz, B., & Vougiouklis, T. (2018). Walk through weak hyperstructures, A: Hv-structures. World Scientific.
Demirci, M. (1999). Fuzzy functions and their fundamental properties. Fuzzy Sets and Systems, 106(2), 239-246.
Du, X. (2025). Multi-Valued Multi-Polar Neutrosophic Sets for Ideological and Political Education Quality in the
New Media Environment. Neutrosophic Sets and Systems, 87, 1042-1052.
Fujita, T. (2025a). Powerset-Theoretic Foundations for HyperAutomata and SuperHyperAutomata. Systemic
Analytics, 3(3), 222-231.
Fujita, T. (2025b). Medical superhyperstructure and healthcare superhyperstructure. Journal of Medicine and
Health Research, 10(2), 243-262.
Fujita, T. (2025c). A theoretical exploration of Hyperconcepts: Hyperfunctions, hyperrandomness, hyperdecision-
making, and beyond (including a survey of hyperstructures). In Fujita, T., & Smarandache, F. (Eds.), Advancing
Uncertain Combinatorics through Graphization, Hyperization, and Uncertainization: Fuzzy, Neutrosophic, Soft,
Rough, and Beyond, vol. 5 (pp. 111-220). NSIA Publishing House.
Fujita, T. (2025d). Short Survey on the Hierarchical Uncertainty of Fuzzy, Neutrosophic, and Plithogenic Sets. In
Fujita, T., & Smarandache, F. (Eds.), Advancing Uncertain Combinatorics through Graphization, Hyperization,
and Uncertainization: Fuzzy, Neutrosophic, Soft, Rough, and Beyond, vol. 4 (pp. 285-314). NSIA Publishing
House.
Fujita, T., Jdid, M., & Smarandache, F. (2025). Hyperfunctions and Superhyperfunctions in Linear Programming:
Foundations and Applications. International Journal of Neutrosophic Science, 26(4), 65-76.
Garcia, M. P., Montiel, O., Castillo, O., Sepulveda, R., & Melin, P. (2009). Path planning for autonomous mobile
robot navigation with ant colony optimization and fuzzy cost function evaluation. Applied Soft Computing, 9(3),
1102-1110.
Hatip, A. (2020). The special neutrosophic functions. International Journal of Neutrosophic Science, 4(2), 104-116.
Huang, M., Zhang, l., & Li, F. (2025). Modeling Cross-Cultural Competence in Vocational Education
Internationalization Using Neutrosophic SuperHyperFunctions and Big Data Driven Cultural Clusters.
Neutrosophic Sets and Systems, 88(1), 362-373.
Jain, M., & Sanga, S. S. (2020). Fuzzy cost optimization and admission control for machine interference problem
with general retrial. Journal of Testing and Evaluation, 48(6), 4488-4502.
Jdid, M., Smarandache, F., & Fujita, T. (2025). A linear mathematical model of the vocational training problem in
a company using neutrosophic logic, hyperfunctions, and SuperHyperFunction. Neutrosophic Sets and Systems,
87, 1-11.
Junco, G. C. A. (2025). Identification of Gaps in Ecuadorian Inclusive University Education through Plithogenic
Offsets. Neutrosophic Sets and Systems, 92, 510-522.
Kumar, T., Kumar, K., Dhiman, N., & Sharma, M. K. (2025). Novel Pythagorean fuzzy score function to optimize
fuzzy transportation models. Results in Engineering, 25, 104048.
Lee, S., & Makhmudov, M. (2024). Decision Making for the Intuitionistic Fuzzy Score Function with the KL-
Divergence on Multi-criteria. In Kahraman, C., Cevik Onar, S., Cebi, S., Oztaysi, B., Tolga, A.C., Ucal Sari, I. (Eds.),
Intelligent and Fuzzy Systems (INFUS 2024). Lecture Notes in Networks and Systems, vol 1088 (pp. 567-574).
Cham: Springer.
Leoreanu-Fotea, V., Rosenberg, I., Davvaz, B., & Vougiouklis, T. (2015). A new class of n-ary hyperoperations.
European Journal of Combinatorics, 44, 265-273.
Nancy, & Garg, H. (2016). An improved score function for ranking neutrosophic sets and its application to decision-
making process. International Journal for Uncertainty Quantification, 6(5), 377-385.
227
Journal of Decision Analytics and Intelligent Computing 5(1) (2025) 219-228 Fujita and Mehmood
Naveed, H., & Ali, S. (2024). Multi-Criteria Decision-Making Approach Based on Correlation Coefficient for Multi-
Polar Interval-Valued Neutrosophic Soft Set. Neutrosophic Systems with Applications, 24, 18-33.
Perfilieva, I. (2004). Fuzzy function as an approximate solution to a system of fuzzy relation equations. Fuzzy sets
and systems, 147(3), 363-383.
Rajalakshmi, R., Lakshmi, M. G., Donganont, M., Raut, P. K., & Mohanty, B. S. (2025). Application of Complex
Neutrosophic Sets to Real-World Decision Making Problems. Neutrosophic Sets and Systems, 91, 209-221.
Saini, R. K., Smarandache, F., Kasim, M., & Ahirwar, A. (2025). Decision-Making Modeling in Agro-Food Systems
Using Type-2 Interval-Valued Linguistic Complex Neutrosophic Sets. Neutrosophic Sets and Systems, 87, 948-
969.
Smarandache, F. (2017). HyperUncertain, SuperUncertain, and SuperHyperUncertain
Sets/Logics/Probabilities/Statistics. Infinite Study.
[Link]
J&redir_esc=y, Accessed 23 May 2025.
Smarandache, F. (2018). Plithogenic set, an extension of crisp, fuzzy, intuitionistic fuzzy, and neutrosophic sets-
revisited. Neutrosophic Sets and Systems, 21, 153-166.
Smarandache, F. (2022). The SuperHyperFunction and the Neutrosophic SuperHyperFunction (revisited again).
Neutrosophic Sets and Systems, 49, 594-600.
Smarandache, F. (2024). Foundation of SuperHyperStructure & Neutrosophic SuperHyperStructure. Neutrosophic
Sets and Systems, 63, 367-381.
Spartalis, S. (1996). On the number of Hv-rings with P-hyperoperations. Discrete Mathematics, 155(1-3), 225-231.
Tak, N. (2020). Type-1 recurrent intuitionistic fuzzy functions for forecasting. Expert Systems with Applications,
140, 112913.
Vadivel, A., & Sundar, C. J. (2021). Application of neutrosophic sets based on mobile network using neutrosophic
functions. 2021 Emerging Trends in Industry 4.0 (ETI 4.0) (pp. 1-8). Raigarh: IEEE.
Vasant, P. M., Andeeski, C. J., Barsoum, N. N., Kolemisevska, T. D., & Dimirovski, G. M. (2006). A fuzzy cost function
optimization in product mix selection problem. 2006 World Automation Congress (pp. 1-6). Budapest: IEEE.
Vougiouklis, T. (2005). A hyperoperation defined on a groupoid equipped with a map. Ratio Mathematica, 15(1),
25-36.
Wang, H., Smarandache, F., Zhang, Y., & Sunderraman, R. (2010). Single valued neutrosophic sets. In Smarandache,
F. (Ed.), Multispace & Multistructure. Neutrosophic Transdisciplinarity (100 Collected Papers of Sciences), vol
IV (pp. 410-413). Hanko: North-European Scientific Publishers.
Wang, X., & Yang, L. (2025). Probabilistic Single-Valued Neutrosophic Set for A Multidimensional Evaluation of
College Students’ Mental Health and Academic Success. Neutrosophic Sets and Systems, 91, 928-936.
Wu, L. (2025). Multipolar Interval-Valued Neutrosophic Soft Set for Integrating Sustainability into Logistics: A
Performance-Based Evaluation of Green Supply Chains. Neutrosophic Sets and Systems, 88, 988-998.
Xu, D., Peng, Y., Hu, L., & Jiang, L. (2025). A Multi-Attribute Group Decision-Making Approach Based on Entropy of
Single-Valued Neutrosophic Set for TOPSIS. Engineering Letters, 33(8), 3129-3138.
Zadeh, L. A. (1965). Fuzzy sets. Information and control, 8(3), 338-353.
Zheng, J., & Wang, Y. (2025). Neutrosophic SuperHyperfunction Invariants for Teaching Quality in College Big Data
Programs. Neutrosophic Sets and Systems, 93, 397-408.
228