You are on page 1of 14

Soft Computing

https://doi.org/10.1007/s00500-020-05180-4

METHODOLOGIES AND APPLICATION

Pragmatic results in Taiwan education system based IVFG & IVNG


Soumitra Poulik1 · Ganesh Ghorai1 · Qin Xin2

© Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
An interval-valued fuzzy graph (IVFG) and degree of vertices have been applied for performance evaluation in an educational
system. The approach is mainly developed based on real membership values of vertices of an IVFG. In this manuscript,
some definitions of generalized fuzzy graphs and neutrosophic graphs structures are improved. First, we have concentrated
to improve the existing definitions for union and join of two IVFG’s, complete IVFG with supporting examples. Then, their
modified version is developed. Secondly, the modified version of interval-valued neutrosophic graph (IVNG) is given. Third,
an algorithm and a flowchart of the proposed method are described. Fourth, the generalized form of complete and strong
IVNG is given with examples. Finally, a real-life application using interval-valued fuzzy graph in education system in Taiwan
is exhibited.

Keywords Interval-valued fuzzy graphs · Union and join · Interval-valued neutrosophic graphs · Complete and strong IVNG ·
Education system

Abbreviations μhB Upper bound of membership value of edges


FG Fuzzy graph in IVFG
IVFG Interval-valued fuzzy graph Sl , I l , Dl Lower bounds of membership, neutral &
IVNG Interval-valued neutrosophic graph non-membership values of vertices of IVNG,
TWAEA Taiwan Assessment and Evaluation Associa- respectively
tion Su , I u , Du Upper bounds of membership, neutral & non-
μlA Lower bound of membership value of vertex membership values of vertices of IVNG,
in IVFG respectively
μhA Upper bound of membership value of vertex Ŝl , Iˆl , D̂l Lower bounds of membership, neutral &
in IVFG non-membership values of edges of IVNG,
μB
l Lower bound of membership value of edge in respectively
IVFG Ŝ u , Iˆu , D̂ u Upper bounds of membership, neutral &
non-membership values of edges of IVNG,
respectively
Communicated by V. Loia.

B Ganesh Ghorai
math.ganesh@mail.vidyasagar.ac.in 1 Introduction
Soumitra Poulik
poulikmsoumitra@gmail.com 1.1 Research background
Qin Xin
qinx@setur.fo Graph theory and its various operations have a huge num-
1 ber of applications for solving many problems which are
Department of Applied Mathematics with Oceanology and
Computer Programming, Vidyasagar University, Midnapore connected to our real life. Xin and Xia (2017) explained
721 102, India application of graph for optimal broadcasting in wireless net-
2 Faculty of Science and Technology, University of the Faroe works. Algorithm and wireless network in graph theory are
Islands, Vestarabryggja 15, FO 100 Tórshavn, Faroe Islands clearly described in Manne and Xin (2008), Yu et al. (2014).

123
S. Poulik et al.

Zadeh’s fuzzy sets (Zadeh 1965) gives the modern mathe- ship, non-membership and neutral value, then single-valued
matics an extraordinary tool to explain the situation where neutrosophic graph (Broumi et al. 2016) is more helpful for
uncertainty present. Due to the existence of fuzziness or decision making in these type of situations. Wang et al. (2005)
uncertainty in vertices and edges in graph theory, Rosenfeld’s initiated a clear idea on interval-valued neutrosophic sets.
fuzzy graph (FG) (Rosenfield 1975) helps us in decision mak- Zhang et al. (2014) described applications of interval-valued
ing in many application field including optimization, network neutrosophic logic in decision making. Broumi et al. (2014)
routing, computer engineering, artificial intelligence, image defined relation based on interval-valued neutrosophic sets.
segmentation, city planning, medical science, etc. The strong Again, if the three characteristics are moving in a fixed range,
path between any two vertices in FG is formulated in Bhutani i.e., lies in an interval, then IVNG can be used to get more
and Rosenfeld (2003). Many operations and their properties precious for decision making.
in FG theory have been clearly explained in Mordeson and
Peng (1994). Ghorai and Pal (2016, 2017, 2018) established 1.2 Framework of this study
various type of FG with their several properties. Different
types of nodes and indices in FG have been discussed in This paper is structured as follows: Sect. 1.3 is based on
Poulik and Ghorai (2020a), Poulik and Ghorai (2020b). the description and comparison of some related works. Sec-
Cornelis et al. (2006) introduced the concept of interval- tion 2 contains some basic definitions which are used in this
valued fuzzy sets. Bustince and Burillo (2000) initiated manuscripts. Section 3 contains some defects in the defini-
interval-valued fuzzy relation and its application in approx- tion of union, join of two IVFGs and complete IVFG. These
imation reasoning. Guha et al. (2009) introduced clustering are discussed by example here and their corrected version
which are based on interval-valued fuzzy relation. Mezei and are established. In Sect. 4, we have discussed the definition
Wikstrom (2013) defined many operators based on interval- of IVNG and showed by example that this definition is inap-
valued fuzzy numbers. In some connected, we have seen that propriate. The corrected definition is given with example.
the membership values of vertices and edges in FG are not Section 5 contains algorithm and flowchart of our proposed
constant and these varies in a range whose upper and lower method. Also, by this definition of IVNG, we found some
values are fixed. The IVFG (Akram and Dudek 2011) gives defect in the definitions of complete and strong IVNG and
more errorless solution in this situation. After that, Jan et al. then their updated versions are given defined. In Sect. 6, an
(2019) created the generalized IVFG and its several proper- application of IVFG is provided to determine the order of the
ties which are used for solving many problems in real-life main criteria’s in Taiwan education system. Some hypothe-
fields like wireless network, CCTV network system, water sis, limitations and advantages, disadvantages are discussed
connection planning in town, decision making in education in Sects. 7 and 8, respectively. Lastly, conclusion is given in
system, etc. Talebi et al. (2016) explained the operations Sect. 9.
on IVFG. Many works depending on data mining analytic
had been done in Ali (2013), Al-Janabi et al. (2020), Mahdi 1.3 Related works
and Al-Janabi (2020). In Al-Janabi (2020), Alkaim and Al-
Janabi (2020) optimization problems in education system and Rosenfield (1975) first gives the notion of fuzzy graph and
gas flaring reduction are discussed. Al-Janabi and Alkaim this idea can be used if the vertices and edges are uncertain
(2020), Al-Janabi et al. (2020) introduced many mathemati- with membership values lies in [0, 1]. But, if the membership
cal models on innovative synthesis and collaborative analysis values are not fixed, i.e., lies in a subinterval of [0, 1], then
of renewable energy and prediction for decision making with IVFG can be used to measure the uncertainty of vertices and
algorithms. In Al-Janabi et al. (2020), using recurrent neu- edges.
ral network, an intelligent predictor model for air pollutants Akram and Dudek (2011) defined IVFG and its various
based on deep learning techniques are designed. operations like union, join, etc. But, in this work the definition
Considering the neutral opinion and abstaining informa- of IVFG, join and union of two IVFG are not true in general.
tion for the analysis, Wang et al. (2012) structured a form The modified definition defined in Jan et al. (2019) and the
of neutrosophic logic and neutrosophic sets. Peng and Dai operations join, union of two IVFG has been established in
(2018) explained many application of interval-valued neu- our work.
trosophic logic in decision making and similarity measure Jan et al. (2019) introduced many modified definitions of
for score function. Yang et al. (2016), Yang et al. (2017) IVFG, IVNG, complete IVFG and their complement. But, the
introduced different types of interval-valued sets and rela- definition of IVFG is not true in general. Here, the definition
tion with their properties. If the vertices or edges or both of IVNG has been corrected with examples.
in a connected graph, work to help the system or does not In Poulik and Ghorai (2020c) and Rashmanlou and Jouy-
any work or works against the system means the vertices and bari (2017), degree of vertices and edges in an IVFG and
edges have three characteristic values which are member- bipolar fuzzy graphs (membership values having two oppo-

123
Pragmatic results in Taiwan education system based IVFG & IVNG

Table 1 Related works of IVFG and IVNG


Authors Years Works

Rosenfield (1975) 1975 Generalized fuzzy graphs


Guha et al. (2009) 2009 Interval-valued fuzzy relation-based clustering with its application to
performance evaluation
Akram and Dudek (2011) 2011 Interval-valued fuzzy graphs
Rashmanlou and Jouybari (2017) 2017 New concepts of interval-valued fuzzy graphs with application
Jan et al. (2019) 2019 Some root level modifications in interval valued fuzzy graphs and their
generalizations including neutrosophic graphs
Al-Janabi and Alkaim (2020) 2020 A nifty collaborative analysis to predicting a novel tool (DRFLLS) for
missing values estimation
Al-Janabi et al. (2020) 2020 An innovative synthesis of deep learning techniques (DCapsNet & DCOM)
for generation electrical renewable energy from wind energy
Al-Janabi et al. (2020) 2020 A new method for prediction of air pollution based on intelligent computation
This paper – Pragmatic results in Taiwan education system based IVFG & IVNG

site components positive and negative). But, a real-life Definition 2.4 (Mordeson and Peng 1994) The join of two
application of degree of vertices in an IVFG is presented FGs G  = (V  , μ , σ  ) and G  = (V  , μ , σ  ) of the graphs
∗ ∗
in this work. G  = (V  , E  ) and G  = (V  , E  ) with V  ∩ V  = φ is
Guha et al. (2009) explained the idea of Taiwan education denoted by G  + G  = (V  ∪ V  , μ + μ , σ  + σ  ) and is
system in terms of interval-valued fuzzy relation. In our work, defined as
all the criteria are shown by an IVFG and the main important
criteria are determined (Table 1). (i) (μ + μ )(a) = (μ ∪ μ )(a), ∀a ∈ V  ∪ V 
(ii) (σ  + σ  )(ab) = (σ  ∪ σ  )(ab), if ab ∈ E  ∪ E 
(iii) (σ  + σ  )(ab) = μ (a) ∧ μ (b), if ab ∈ E 1 , a ∈
2 Preliminary V  , b ∈ V  where E 1 is the set of all edges joining the
nodes of V  and V  .
Here, some basic definitions related to FG and IVFG are
presented, which have been used in this paper. Definition 2.5 (Jan et al. 2019) An IVFG G is a pair (A, B)
of the graph G ∗ = (V , E), where V and E are the set of all
Definition 2.1 (Rosenfield 1975) A FG G = (V , μ, σ ) of nodes and edges of G, respectively, satisfying the following
the graph G ∗ = (V , E) is defined by μ : V → [0, 1] and two conditions
σ : V × V → [0, 1] such that for all a, b ∈ V ,
σ (ab) ≤ μ(a) ∧ μ(b), where μ(a) and σ (ab) values of the (i) for any b ∈ V , the function μ A (b) represents the
vertex a and the edge ab in G, respectively, and ‘∧’ denote membership value of b and
minimum value.
μ A (b) = [μlA (b), μhA (b)] is a subinterval of [0, 1],
Definition 2.2 (Mordeson and Peng 1994) A FG G is com-
plete if σ (ab) = μ(a) ∧ μ(b), ∀a, b ∈ V .
(ii) for any ab ∈ E, the function μ B (ab) represents the
Definition 2.3 (Mordeson and Peng 1994) The union of two membership value of ab and
FGs G  = (V  , μ , σ  ) and G  = (V  , μ , σ  ) of the graphs μ B (ab) = [μlB (ab), μhB (ab)] is a subinterval of [0, 1]
∗ ∗
G  = (V  , E  ) and G  = (V  , E  ) is denoted by G  ∪ such that
G  = (V  ∪ V  , μ ∪ μ , σ  ∪ σ  ) and is defined as μlB (ab) ≤ min{μlA (a), μlA (b)} and
μhB (ab) ≤ min{μhA (a), μhA (b)} with μhB (ab) ≥ min
(i) (μ ∪ μ )(a) = μ (a) if a ∈ V  − V  , {μlA (a), μlA (b)}.
(μ ∪ μ )(a) = μ (a) if a ∈ V  − V  and
(μ ∪ μ )(a) = max{μ (a), μ (a)} if a ∈ V  ∩ V  , First, we recall the definition of the union of two IVFGs
(ii) (σ  ∪ σ  )(ab) = σ  (ab), if ab ∈ E  − E  , given in Akram and Dudek (2011). It is shown in Sect. 3 that
(σ  ∪ σ  )(ab) = σ  (ab), if ab ∈ E  − E  and this definition is not true in general.
(σ  ∪ σ  )(ab) = max{σ  (ab), σ  (ab)}, if ab ∈ E  ∩ E  . Definition 2.6 (Akram and Dudek 2011) The union of two
IVFGs G  = (A , B  ) and G  = (A , B  ) of the graphs

123
S. Poulik et al.

∗ ∗
G  = (V  , E  ) and G  = (V  , E  ) is denoted by G  ∪ Example 3.1 (Example 3.10 of Akram and Dudek 2011)
G  = (A ∪ A , B  ∪ B  ) and is defined by Here, we write G  , G  , V  , V  , E  , E  , A , A , B  , B  in
place of G 1 , G 2 , V1 , V2 , E 1 , E 2 , A1 , A2 , B1 , B2 , respectively.

⎪ 
⎨(μ A ∪ μ A )(a) = μ A (a) if a ∈ V and a ∈
l l l / V  Here, G  and G  satisfy the Definition 2.5. So G  and G  are
(i) (μ A ∪ μ A )(a) = μ A (a) if a ∈
l l l / V and a ∈ V 
 IVFGs. But in G  ∪ G  , we see that

⎩ l
(μ A ∪ μ A )(a) = max{μ A (a), μlA (a)} if a ∈ V  ∩ V  .
l l (μlA ∪ μlA )(b) = 0.4, (μhA ∪ μhA )(b) = 0.5, (μlA ∪
⎧ μ A )( f ) = 0.4, (μhA ∪ μhA )( f ) = 0.6,
l
⎪  / V 
⎨(μ A ∪ μ A )(a) = μ A (a) if a ∈ V and a ∈
h h h
(μlB  ∪ μlB  )(b f ) = 0.1, (μhB  ∪ μhB  )(b f ) = 0.2.
(ii) (μ A ∪ μ A )(a) = μ A (a) if a ∈
h h h / V and a ∈ V 


⎩ h Here (μhB  ∪μhB  )(b f ) = 0.2  min{(μlA ∪μlA )(b), (μlA ∪
(μ A ∪ μhA )(a) = max{μhA (a), μhA (a)} if a ∈ V  ∩ V  ,
⎧ μ A )( f )} = min{0.4, 0.4} = 0.4, which contradict the fact
l
⎪  / E 
⎨(μ A ∪ μ A )(ab) = μ B  (ab) if ab ∈ E and ab ∈
l l l
that the union of two IVFGs is again an IVFG, by Proposition
(iii) (μ A ∪ μ A )(ab) = μ B  (ab) if ab ∈
l l l / E and ab ∈ E 

3.11 of Akram and Dudek (2011).

⎩ l
(μ A ∪ μ A )(ab) = max{μ B  (ab), μlB  (ab)} if ab ∈ E  ∩ E  ,
l l
⎧ So the Definition 3.9 of Akram and Dudek (2011) is incor-
⎪  / E 
⎨(μ A ∪ μ A )(ab) = μ B  (ab) if ab ∈ E and ab ∈
h h h

(iv) (μ A ∪ μ A )(ab) = μ B  (ab) if ab ∈


h h h / E and ab ∈ E 
 rect. The modified version is established in such a way that

⎩ h Proposition 3.11 of Akram and Dudek (2011) becomes true.
(μ A ∪ μ A )(ab) = max{μ B  (ab), μhB  (ab)} if ab ∈ E  ∩ E  .
h h

Definition 3.2 (Modified definition of union of two IVFGs)


Now, recall the definition of join of two IVFGs given in ∗
The union of two IVFGs G  and G  of the graphs G  and
(Akram and Dudek 2011) and in Sect. 3 we will show by ∗
G  is denoted by G  ∪ G  = (A ∪ A , B  ∪ B  ) defined by
counter example that this definition is also not correct.

l l l  / V 
Definition 2.7 (Akram and Dudek 2011) The join of two ⎨(μ A ∪ μ A )(a) = μ A (a) if a ∈ V and a ∈

∗  and a ∈ V 
IVFGs G  and G  of the graphs G  and (i) (μl  ∪ μl  )(a) = μl  (a) if a ∈

/ V
∗ ⎩ lA A A
G  denoted by G  + G  = (A + A , B  + B  ) and is (μ A ∪ μlA )(a) = max{μlA (a), μlA (a)} if a ∈ V  ∩ V  .
⎧ h
defined as ⎪
⎪ (μ ∪ μhA )(a) = μhA (a) if a ∈ V  and a ∈ / V 
⎨ A
 (ii) (μ  ∪ μ  )(a) = μ  (a) if a ∈
h h h / V and a ∈ V 


⎪ A A A
(μlA + μlA )(a) = (μlA ∪ μlA )(a) ⎩ h
(i) (μ A ∪ μhA )(a) = max{μhA (a), μhA (a)} if a ∈ V  ∩ V  ,
(μhA + μhA )(a) = (μhA ∪ μhA )(a) if a ∈ V  ∪ V  , ⎧ l
 ⎪
⎪ (μ ∪ μlA )(ab) = min{(μlA ∪ μlA )(a), (μlA ∪ μlA )(b)
(μlB  + μlB  )(ab) = (μlB  ∪ μlB  )(ab) ⎨ A
(ii) (iii) if either ab ∈ E  or ab ∈ E  but ab ∈ / E  ∩ E  ,
(μhB  + μhB  )(ab) = (μhA ∪ μhA )(ab) if x y ∈ E  ∩ E  , ⎪

⎩ l
 (μ A ∪ μlA )(ab) = max{μlB  (ab), μlB  (ab)} if ab ∈ E  ∩ E  ,
(μlB  + μlB  )(ab) = min(μlA (a), μlA (b)) ⎧
(iii) h h h h h h
⎨(μ A ∪ μ A )(ab) = min{(μ A ∪ μ A )(a), (μ A ∪ μ A )(b)

(μhB  + μhB  )(ab) = min(μhA (a), μhA (b)) if ab ∈ E,    
(iv) if either ab ∈ E or ab ∈ E but ab ∈ / E ∩E ,

⎩ h
(μ A ∪ μhA )(ab) = max{μhB  (ab), μhB  (ab)} if ab ∈ E  ∩ E  .
where E is the set of all edges joining the nodes of V  and
V  . Example 3.3 From Example 3.1, G  and G  of Fig. 1 are
IVFGs. Now according to the Definition 3.2, we have

3 Operations on IVFG (μlA ∪ μlA )(b) = 0.4,

In FG, the membership values of the vertices as well as edges (μlA ∪ μlA )(b) = 0.5,
lie in a subinterval of [0, 1]. Then, the IVFG and its vari- (μlA ∪ μlA )( f ) = 0.4,
ous operations give better results for solving many real-life (μlA ∪ μlA )( f ) = 0.6,
problems. In this section, first the definition of union of two (μlB  ∪ μlB  )(b f ) = 0.4,
IVFGs, join of two IVFGs and complete IVFG are reviewed
(μlB  ∪ μlB  )(b f ) = 0.5.
by various counterexamples. After that, the new generalized
form of all these definitions is developed with examples. 0.4 ≤ min{0.4, 0.4}, 0.5
Throughout the paper, G ∗ denotes the crisp graph and G ≤ min{0.5, 0.6}, with0.5 ≥ min{0.4, 0.4}
denotes the IVFG of G ∗ . ⇒ (μlB  ∪ μlB  )(b f ) ≤ min{(μlA ∪ μlA )(b),
Here, we consider the example given in Akram and Dudek
(μlA ∪ μlA )( f )},
(2011). Then, we show that this example which is constructed
based on the Definition 2.6 does not satisfy the definition of (μhB  ∪ μhB  )(b f ) ≤ min{(μhA ∪ μhA )(b),
IVFG 2.5. (μhA ∪ μhA )( f )}with

123
Pragmatic results in Taiwan education system based IVFG & IVNG

Fig. 1 Two IVFGs G  , G  and their union G  ∪ G 

μlB  (ab) = 0.1 ≤ min{μlA (a), μlA (b)} and


μhB  (ab) = 0.2 ≤ min{μhA (a), μhA (b)} with μhB  (ab) ≥
min{μlA (a), μlA (b)}.
So G  is an IVFG. Similarly, G  is also an IVFG.
But to determine G  + G  , using Definition 2.7, we can-
not determine (μlB  + μlB  )(ab), (μhB  + μhB  )(ab), (μlB  +
μlB  )(ad) and (μhB  + μhB  )(ad), since a ∈ V  ∩ V  , b ∈ V  ,
b ∈ V  and it is not given in Definition 3.12 of Akram and
Dudek (2011). So G  + G  is undefined. Thus, the above def-
Fig. 2 Join G  + G  of two IVFGs G  , G  using Definition 2.7 inition of join of two IVFGs in Akram and Dudek (2011) is
incorrect. Now, we develop a definition of join of two IVFGs
to overcome this type of situation.
(μhB  ∪ μhB  )(b f ) ≥ min{(μlA ∪ μlA )(b),
(μlA ∪ μlA )( f )}. The modified definition of join of two IVFG is defined as
follows.
So the edge b f of G  ∪ G  satisfies the conditions of Defini- Definition 3.5 (Modified definition of join of two IVFGs) The
tion 3.2 of an IVFG. Similarly, all the other edges of G  ∪ G  ∗
join of two IVFGs G  and G  of the graphs G  = (V  , E  )
also satisfy the conditions of Definition 3.2 of an IVFG. ∗
and G  = (V  , E  ) with V  ∩ V  = φ is denoted by
Hence, G  ∪ G  is an IVFG. G  + G  = (A + A , B  + B  ) and is defined as
Now, we consider an example to construct join of two 
(μlA + μlA )(a) = (μlA ∪ μlA )(a)
IVFGs using Definition 2.7. Then, show that this definition (i)
(μhA + μhA )(a) = (μhA ∪ μhA )(a) if a ∈ V  ∪ V  ,
does not satisfy the Definition 2.5 of IVFG . 
(μlB  + μlB  )(ab) = (μlB  ∪ μlB  )(ab)
Example 3.4 Consider the IVFGs G  and G  and their join (ii)
(μhB  + μhB  )(ab) = (μhA ∪ μhA )(ab) if x y ∈ E  ∩ E  ,
G  + G  as shown in Fig. 2. 
Here, μlA (a) = 0.1, μhA (a) = 0.2, μlA (b) = 0.3, μhA (b) = (μlB  + μlB  )(ab) = min(μlA (a), μlA (b))
(iii)
0.4, (μhB  + μhB  )(ab) = min(μhA (a), μhA (b)) if ab ∈ E,

123
S. Poulik et al.

Fig. 4 An IVFG G

Fig. 3 Two IVFGs G  , G  and their join G  + G 

where E is the set of all edges joining the nodes of V1 and


V2 .

Example 3.6 Consider the IVFG G  and G  of Fig. 3. Here,


V  = {a, b}, V  = {c, d}, so V  ∩ V  = φ.
Fig. 5 A complete IVFG G
μlA (a) = 0.1, μhA (a) = 0.2, μlA (b) = 0.3, μhA (b) = 0.4,
μlA (c) = 0.4, μhA (c) = 0.5, μlA (d) = 0.5, μhA (d) = 0.8,
μlB  (ab) = 0.1, μhB  (ab) = 0.2, μlB  (cd) = 0.4, μhB  (cd) = μlB (ab) = min{μlA (a), μlA (b)} and μhB (ab) = min{μhA
0.5. (a), μhA (b)}, ∀ab ∈ E.
According to Definition 3.5, we have
Example 3.8 For the IVFG G of Fig. 4, we see that there is no
(μlA + μlA )(a) = 0.1, (μhA + μhA )(a) = 0.2, edge between the nodes a and b. So, [μlB (ab), μhB (ab)] =
(μlA + μlA )(b) = 0.3, (μhA + μhA )(b) = 0.4, [0, 0] = [0.3, 0.7] = [min{μlA (a), μlA (b)}, min{μhA (a),
μhA (b)}]. So, here G is a strong IVFG but not a complete
(μlA + μlA )(c) = 0.4, (μhA + μhA )(c) = 0.5,
IVFG. Thus, the above definition is incorrect.
(μlA + μlA )(d) = 0.5, (μhA + μhA )(d) = 0.8,
(μlB  + μlB  )(ab) = 0.1, (μhB  + μhB  )(ab) = 0.2, The modified definition of complete IVFG is defined
below.
(μlB  + μlB  )(ac) = 0.1, (μhB  + μhB  )(ac) = 0.2,
(μlB  + μlB  )(cd) = 0.4, (μhB  + μhB  )(cd) = 0.5, Definition 3.9 (Modified definition of complete IVFG) An
IVFG G = (A, B) is said to be complete if
(μlB  + μlB  )(bd) = 0.3, (μhB  + μhB  )(bd) = 0.4,
μlB (ab) = min{μlA (a), μlA (b)} and μhB (ab) = min{μhA
(μlB  + μlB  )(ad) = 0.1, (μhB  + μhB  )(ad) = 0.2, (a), μhA (b)}, ∀a, b ∈ V .
(μlB  + μlB  )(bc) = 0.3, (μhB  + μhB  )(bc) = 0.4.
Example 3.10 For the IVFG G of Fig. 5, we have
Now, 0.1 ≤ min{0.1, 0.3}, 0.2 ≤ min{0.2, 0.4} with 0.2 ≥ μlA (a) = 0.5, μhA (a) = 0.8, μlA (b) = 0.3, μhA (b) = 0.7,
min{0.1, 0.3} μlA (c) = 0.2, μhA (c) = 0.7, μlB (ab) = 0.3, μhB (ab) = 0.7,
μlB (bc) = 0.2, μhB (bc) = 0.7, μlB (ac) = 0.2, μhB (ac) =
⇒ (μlB  ∪ μlB  )(ab) ≤ min{(μlA ∪ μlA )(a), (μlA ∪ μlA )(b)} 0.7.
(μhB  ∪ μhB  )(ab) ≤ min{(μhA ∪ μhA )(a), (μhA ∪ μhA )(b)} with
So μlB (ab) = min{μlA (a), μlA (b)}, μhB (ab) = min{μhA (a),
μ A (b)} with μhB (ab) = min{μlA (a), μlA (b)}.
h
(μhB  ∪ μhB  )(ab) ≥ min{(μlA ∪ μlA )(a), (μlA ∪ μlA )(b)}.
So the edge ab of G satisfies the conditions of a complete
IVFG. Similarly, all the other edges of G also satisfy the
So the edge ab of G  + G  satisfies the conditions of Defini-
conditions of Definition 3.9 of a complete IVFG. Therefore,
tion 2.4 of an IVFG. Similarly, all the other edges of G  + G 
G is a complete IVFG. Hence, G is a complete IVFG.
also satisfy the conditions of Definition 2.4 of an IVFG.
Hence, G  + G  is an IVFG.

Now, we recall the definition of complete IVFG given in 4 Interval-valued neutrosophic graph
Akram and Dudek (2011) and verify its flaws in the following
example. In various problems connected to graph theory, it is seen
that sometimes the vertices behave as a membership and
Definition 3.7 An IVFG G = (A, B) is said to be complete sometimes its constant means neutral and sometimes its
if work against the system means non-membership. So, the

123
Pragmatic results in Taiwan education system based IVFG & IVNG

vertices have three characters and similarly every edges also


have three characters. All these three character cannot work
together for all the vertices and edges, so characters are uncer-
tain and lies in interval. Therefore, to control many conditions Fig. 6 An IVNG G
of this form, IVNG plays one of the egregious role. In this sec-
tion, we pointed out many flaws in the definitions of IVNG,
complete IVNG and strong IVNG. The generalized version denoted by S, I and D, respectively. Mainly, S = [Sl , S u ],
of all these definitions are established. I = [I l , I u ], D = [Dl , D u ] and all of these are subintervals
First, let us recall the definition of IVNG given in Jan et al. of [0, 1] such that 0 ≤ S u + I u + D u ≤ 3;
(2019) and we will show that this definition is incorrect by (ii) for all edges ai a j ∈ E, there are three functions
giving an example. namely membership, neutral and non-membership value of
ai a j ∈ V × V which are denoted by Ŝ, Iˆ and D, respectively.
Definition 4.1 (Jan et al. 2019) A pair G = (V , E), where V Mainly, Ŝ = [ Ŝl , Ŝ u ], Iˆ = [ Iˆl , Iˆu ] and D̂ = [ D̂l , D̂ u ] are
denote the set of all nodes and E denote the set of all edges, subinterval of [0, 1] such that
is called an IVNG if the following are satisfied
(i) for all ai ∈ V , there are three functions namely Ŝl (ai , a j ) ≤ min(Sl (ai ), Sl (a j )),
the membership, neutral and non-membership value of ai ,
Ŝ u (ai , a j ) ≤ min(S u (ai ), S u (a j ))
which are denoted by S, I and D, respectively. Mainly,
S = [Sl , S u ], I = [I l , I u ], D = [Dl , D u ] and all of these Iˆl (ai , a j ) ≤ max(I l (ai ), I l (a j )),
are subintervals of [0, 1] such that 0 ≤ S u + I u + D u ≤ 3; Iˆu (ai , a j ) ≤ max(I u (ai ), I u (a j ))
(ii) for all edges ai a j ∈ E, there are three functions
D̂l (ai , a j ) ≤ max(Dl (ai ), Dl (a j )),
namely membership, neutral and non-membership value of
ai a j ∈ V × V which are denoted by Ŝ, Iˆ and D, respec- D̂ u (ai , a j ) ≤ max(D u (ai ), D u (a j ))
tively. Mainly, Ŝ = [ Ŝl , Ŝl ], Iˆ = [ Iˆl , Iˆu ] and D̂ = [ D̂l , D̂l ]
are subinterval of [0, 1] such that provided that Ŝl (ai , a j ) ≤ Ŝ u (ai , a j ), Iˆl (ai , a j ) ≤ Iˆu (ai , a j ),
D̂l (ai , a j ) ≤ D̂ u (ai , a j ) and 0 ≤ Ŝ u + Iˆu + D̂ u ≤ 3.
Ŝl (ai , a j ) ≤ min(Sl (ai ), Sl (a j )),
Now, we show by example that the Definition 4.2 also vio-
Ŝ u (ai , a j ) ≤ min(S u (ai ), S u (a j )) lates the definition of complement of an IVNG. For this, first
Iˆl (ai , a j ) ≤ max(I l (ai ), I l (a j )), we recall the Definition 18 of Jan et al. (2019) of complement
Iˆu (ai , a j ) ≤ max(I u (ai ), I u (a j )) of an IVNG.

D̂l (ai , a j ) ≤ max(Dl (ai ), Dl (a j )), Definition 4.3 (Jan et al. 2019) The complement of an IVNG
D̂ (ai , a j ) ≤ max(D (ai ), D (a j ))
u u u G = (V , E) is defined by G c = (V c , E c ) where V c = V
and the membership, abstinence and non-membership grades
provided that Ŝ u (ai , a j ) ≤ Ŝl (ai ,a j ), Iˆl (ai , a j ) ≤ Iˆu (ai , a j ), of E satisfy the conditions:
D̂l (ai , a j ) ≤ D̂ u (ai , a j ) and 0 ≤ Ŝ u + Iˆu + D̂ u ≤ 3.
( Ŝl )c (ai , a j ) = min(Sl (ai ), S L (a j )) − Ŝl ((ai , a j ))
There are two types of mistakes in the Definition 4.1. ( Ŝ u )c (ai , a j ) = min(S u (ai ), S u (a j )) − Ŝ u (ai , a j )
First there are typing errors and in the second place the con- + min(Sl (ai ), Sl (a j ))
ditions “ Ŝ u (ai , a j ) ≤ Ŝl (ai , a j ), Iˆl (ai , a j ) ≤ Iˆu (ai , a j ),
( Iˆl )c (ai , a j ) = max(I l (ai ), I l (a j )) − Iˆl ((ai , a j ))
D̂l (ai , a j ) ≤ D̂ u (ai , a j )” have to be modified.
If we use this definition of IVNG, then the complement ( Iˆu )c (ai , a j ) = max(I u (ai ), I u (a j )) − Iˆu (ai , a j )
of an IVNG (Definition 18 of Jan et al. 2019) may not be an + max(I l (ai ), I l (a j ))
IVNG. Now, the typing mistakes are
( D̂l )c (ai , a j ) = max(Dl (ai ), Dl (a j )) − D̂l ((ai , a j ))
Ŝ = [ Ŝl , Ŝl ], D̂ = [ D̂l , D̂l ] and Ŝ u (ai , a j ) ≤ Ŝ u (ai , a j ).
If these typos are corrected, then the Definition 4.1 ( D̂ u )c (ai , a j ) = max(D u (ai ), D u (a j )) − D̂ u (ai , a j )
becomes as follows: + max(Dl (ai ), Dl (a j ))

Definition 4.2 A pair G = (V , E), where V denote the set of Now, we demonstrate by two examples that how the Def-
all nodes and E denote the set of all edges, called an IVNG inition 4.2 of an IVNG violates the above Definition 4.3.
if the following are satisfied
(i) for all ai ∈ V there are three functions namely the mem- Example 4.4 Consider the IVNG G of Fig. 6.
bership, neutral and non-membership value of ai , which are Here, Ŝl (a, b) = 0.1 ≤ min{0.7, 0.6} = min{(Sl (a), Sl (b)},

123
S. Poulik et al.

Fig. 7 G c , the complement of G of Fig. 6

Fig. 9 G c , the complement of G of Fig. 8

(i) for all ai ∈ V , there are three functions namely


the membership, neutral and non-membership value of ai ,
Fig. 8 An IVNG G which are denoted by S, I and D, respectively. Mainly,
S = [Sl , S u ], I = [I l , I u ], D = [Dl , D u ] and all of these
Ŝ u (a, b) = 0.2 ≤ min{0.8, 0.7} = min{(S u (a), S u (b)}. are subintervals of [0, 1] such that 0 ≤ S u + I u + D u ≤ 3;
So Ŝl (a, b) ≤ Ŝ u (a, b). Similarly, Iˆl (a, b) ≤ Iˆu (a, b) (ii) for all edges ai a j ∈ E, there are three functions
and D̂l (a, b) ≤ D̂ u (a, b). namely membership, neutral and non-membership value of
Also, 0 ≤ 0.8+0.9+0.9 ≤ 3, 0 ≤ 0.7+0.8+0.9 ≤ 3 and ai a j ∈ V × V which are denoted by Ŝ, Iˆ and D̂, respectively.
0 ≤ 0.2+0.3+0.4 ≤ 3 means 0 ≤ S u (a)+ I u (a)+ D u (a) ≤ Mainly, Ŝ = [ Ŝl , Ŝ u ], Iˆ = [ Iˆl , Iˆu ] and D̂ = [ D̂l , D̂ u ] are
3, 0 ≤ S u (b)+I u (b)+D u (b) ≤ 3 and 0 ≤ Ŝ u + Iˆu + D̂ u ≤ 3. subinterval of [0, 1] such that

Now, using the Definition 4.3, the complement of the IVNG


Ŝl (ai , a j ) ≤ min(Sl (ai ), Sl (a j )), Ŝ u (ai , a j ) ≤ min(S u (ai ), S u (a j )),
G of Fig. 6 is shown in Fig. 7.
We know that the complement of an IVNG is an IVNG. Ŝ u (ai , a j ) ≥ min(Sl (ai ), Sl (a j )) Iˆl (ai , a j ) ≤ max(I l (ai ), I l (a j )),
But here ( Ŝ u )c (a, b) = 1.1 > 1, ( Iˆu )c (a, b) = 1.2 > 1 and Iˆu (ai , a j ) ≤ max(I u (ai ), I u (a j )), Iˆu (ai , a j ) ≥ max(I l (ai ), I l (a j ))
( D̂ u )c (a, b) = 1.1 > 1. Also, ( Ŝ u )c (a, b) + ( Iˆu )c (a, b) + D̂l (ai , a j ) ≤ max(Dl (ai ), Dl (a j )), D̂ u (ai , a j ) ≤ max(D u (ai ), D u (a j )),
( D̂ u )c (a, b) = 3.4 > 3. So G c is not an IVNG, which con- D̂ u (ai , a j ) ≥ max(Dl (ai ), Dl (a j ))
tradicts that the complement of an IVNG is again an IVNG.

Example 4.5 Consider the IVNG G of Fig. 8 with 0 ≤ Ŝ u + Iˆu + D̂ u ≤ 3.


Here, Ŝl (a, b) = 0.2 ≤ min{0.8, 0.6} = min{(Sl (a), Sl (b)}, In this case, we call G a generalized IVNG.
Ŝ u (a, b) = 0.2 ≤ min{0.9, 0.8} = min{(S u (a), S u (b)}.
If we use above definition for an IVNG, then the comple-
So Ŝl (a, b) ≤ Ŝ u (a, b). Similarly, Iˆl (a, b) ≤ Iˆu (a, b),
ment of an IVNG is always an IVNG. This is verified in the
D̂l (a, b) ≤ D̂ u (a, b), Ŝl (b, c) ≤ Ŝ u (b, c), Iˆl (b, c) ≤
following.
Iˆu (b, c), D̂l (b, c) ≤ D̂ u (b, c).
0 ≤ 0.9 + 0.8 + 0.8 ≤ 3, 0 ≤ 0.8 + 0.8 + 0.8 ≤ 3, 0 ≤ 0.9 + Note 4.7 From Definition 4.6, we have
0.8+0.9 ≤ 3, 0 ≤ 0.3+0.2+0.2 ≤ 3, 0 ≤ 0.2+0.3+0. ≤ 3, 0 ≤ min(Sl (ai ), Sl (a j )) ≤ Ŝ u (ai , a j ) ≤ min(S u (ai ),
means S (a j )) ≤ 1.
u
0 ≤ S u (a) + I u (a) + D u (a) ≤ 3, 0 ≤ S u (b) + I u (b) + Consider min(Sl (ai ), Sl (a j )) = R, Ŝ u (ai , a j ) = Q and
D u (b) ≤ 3, 0 ≤ S u (c) + I u (c) + D u (c) ≤ 3 and 0 ≤ min(S u (ai ), S u (a j )) = P.
Ŝ u + Iˆu + D̂ u ≤ 3. So 0 ≤ R ≤ Q ≤ P ≤ 1, where P, Q, R ∈ [0, 1].
Now, using the Definition 4.3, the complement of the Now, Q ≥ R and Q ≥ R, Q ≤ 1, R ≥ 0
IVNG G of Fig. 8 is shown in Fig. 9. ⇒ Q − R ≥ 0 and Q − R ≤ 1
But here ( Ŝ u )c (a, b) = 1.1 > 1, ( Iˆu )c (a, b) = 1.3 > ⇒ 0 ≤ Q − R ≤ 1.
1 and ( D̂ u )c (a, b) = 1.3 > 1. Also, ( Ŝ u )c (a, b) + Again P ≥ Q, R ≥ 0 ⇒ P ≥ Q − R.
( Iˆu )c (a, b)+( D̂ u )c (a, b) = 3.7 > 3. So G c is not an IVNG, Now, P ≥ Q − R and P ≥ Q − R, P ≤ 1, Q − R ≥ 0
which contradicts that the complement of an IVNG is again ⇒ P − (Q − R) ≥ 0 and P − (Q − R) ≤ 1
an IVNG. ⇒ 0 ≤ P − (Q − R) ≤ 1
⇒ 0 ≤ P − Q + R ≤ 1.
The correct definition of IVNG is given in Definition 4.6.
Thus, 0 ≤ min(S u (ai ), S u (a j ))− Ŝ u (ai , a j )+min(Sl (ai ), Sl ;
Definition 4.6 (Modified definition of IVNG) A pair G = (a j )) ≤ 1
(V , E), where V denote the set of all nodes and E denote the ⇒ 0 ≤ ( Ŝ u )c (ai , a j ) ≤ 1, from Definition 4.3.
set of all edges, is called an IVNG if the following conditions Similarly, it can prove that 0 ≤ ( Iˆu )c (ai , a j ) ≤ 1, 0 ≤
are satisfied ( D̂ u )c (ai , a j ) ≤ 1.

123
Pragmatic results in Taiwan education system based IVFG & IVNG

Hence, according to the Definition 4.8, G is not an IVNG.


Therefore, the Definition 4.8 is not correct. Now, the modified
definition of a complete IVNG is given below with example.

Definition 4.10 (Complete IVNG) An IVNG is called com-


plete if
 
Fig. 10 An IVNG G  1. Ŝl (ai , a j ) = min Sl (ai ), Sl (a j ) ,
 
Ŝ u (ai , a j ) = min S u (ai ), S u (a j ) ,
 
2. Iˆl (ai , a j ) = max I l (ai ), I l (a j ) ,
 
Iˆu (ai , a j ) = max I u (ai ), I u (a j ) ,
 
3. D̂l (ai , a j ) = max Dl (ai ), Dl (a j ) ,
 
D̂ u (ai , a j ) = max D u (ai ), D u (a j ) , for all ai , a j ∈ V .

Now, recall the definition of strong IVNG given in Akram


Fig. 11 G  , complement of G  of Fig. 10
c

and Nasir (2017) and we found out the flaws in this definition
by example.
Definition 4.11 (Akram and Nasir 2017) An IVNG is called
strong if
Fig. 12 A defective complete IVNG  l 
1. Ŝl (ai , a j ) = min  S (ai ), Sl (a j ) , Ŝ u (ai , a j ) = min
S u (ai ), S u (a j ) ,
Therefore 0 ≤ ( Ŝ u )c + ( Iˆu )c + ( D̂ u )c ≤ 3.  l 
2. Iˆl (ai , a j ) = min  I (ai ), I l (a ) , Iˆu (a , a ) = min
j i j
Now, according to Definition 4.6, we consider an example I u (ai ), I u (a j ) ,
 l 
of an IVNG G  which is shown in Fig. 10 and the complement 3. D̂l (ai , a j ) = min  D (ai ), Dl (a j ) , D̂ u (ai , a j ) = min
of G  is G  which is shown in Fig. 11. We can see that (G  )c
c
D u (ai ), D u (a j ) , for all a j a j ∈ E.
is also an IVNG.
Now, recall the definition of complete IVNG given in Example
 l 4.12 Consider the IVNG G of Fig. 14. Here, S(b)  =
Akram and Nasir (2017) and then we found out the flaws S (b), S u (b) = [0.1, 0.2], I (b)  = I l (b), I u (b) =
in this definition by example. [0.3,
 l 0.4],u D(b)  = D (b), D (b) = [0.3,
l u
 0.4] and S(c)  =
S (c), S (c) =  [0.2, 0.3], I (c)  = I l (c), I u (c) =
Definition 4.8 (Akram and Nasir 2017) An IVNG G = [0.1, 0.2], D(c) = Dl (c), D u (c) = [0.1, 0.2].
(V , E) is called complete if Ŝ(b, c) = [0.1, 0.2], Iˆ(b, c) = [0.1, 0.2], D̂(b, c) =
  [0.1, 0.2]. So G satisfies all the conditions of the Defini-
1. Ŝl (ai , a j ) = min  S l (a ), S l (a ) , Ŝ u (a , a ) = min
i j i j tion 4.11.
S u (ai ), S u (a j ) , But, 0.2  max{0.1, 0.3}, i.e., Iˆu (b, c)  max{I l (b), I l (c)}
  
2. Iˆl (ai , a j ) = min I l (ai ), I l (a j ) , Iˆu (ai , a j ) = min I u (ai ), and D̂ u (b, c)  max{Dl (b), Dl (c)}.
I u (a j ) ,
 l  Hence, according to the Definition 4.11, G is not an IVNG.
3. D̂l (ai , a j ) = min  D (ai ), D l (a ) , D̂ u (a , a ) = min
j i j Therefore, the Definition 4.11 is not correct. Now, the modi-
D u (ai ), D u (a j ) , for all ai , a j ∈ V . fied definition of a strong IVNG is given below with example.

Definition 4.13 (Strong IVNG) An IVNG is called strong if


Example 4.9 Consider
 the IVNG G of Fig.12. Here, S(b)  =
 l 
Sl (b), S u (b) = [0.2, 0.3], I (b) = I l (b), I u (b) = 1. Ŝl (ai , a j ) = min S (ai ), Sl (a j ) , Ŝ u (ai , a j ) = min

[0.1,  = D (b), D (b) = [0.1,  =
l u
0.2], D(b)  0.2] and S(c) S u (ai ), S u (a j ) ,
S (c), S (c) =  [0.1, 0.2], I (c)
l u = I (c), I (c) =
l u  l 
 2. Iˆl (ai , a j ) = max  I (ai ), I l (a ) , Iˆu (a , a ) = max
j i j
[0.3, 0.4], D(c) = Dl (c), D u (c) = [0.3, 0.4]. I u (ai ), I u (a j ) ,
Ŝ(b, c) = [0.1, 0.2], Iˆ(b, c) = [0.1, 0.2], D̂(b, c) =  
3. D̂l (ai , a j ) = max Dl (ai ), Dl (a j ) , D̂ u (ai , a j ) =
[0.1, 0.2]. So G satisfies all the conditions of the Defini- max D u (ai ), D u (a j ) , for all ai a j ∈ E.
tion 4.8.
But, 0.2  max{0.1, 0.3}, i.e., Iˆu (b, c)  max{I l (b), I l (c)} Example 4.14 The IVNG G of Fig. 13 is a strong IVNG.
and D̂ u (b, c)  max{Dl (b), Dl (c)}.

123
S. Poulik et al.

Fig. 13 Example of a complete


IVNG

Step 2 Write the membership values μ B (bi b j ) = [μlB


(bi b j ), μhB (bi b j )] ⊆ [0, 1] of the edges bi b j , i = j, 1 ≤
i, j ≤ n such that μlB (bi b j ) ≤ min{μlA (bi ), μlA (b j )},
Fig. 14 A defective strong IVNG μhB (bi b j ) ≤ min{μhA (bi ), μhA (b j )}, with μhB (bi b j ) ≥
min{μlA (bi ), μlA (b j )}.
Step 3 Calculate degree of vertices bi (i = 1, 2, . . . , n),
deg(bi ) = (deg(bi )l , deg(bi )h ), where deg(bi )l = i = j
bi b j ∈E

μlB (bi b j ), deg(bi )h = i = j μ B (bi b j ).
h
bi b j ∈E
Step 4 Calculate half of sum of the upper and lower bound-
deg(b )l +deg(b )h
i j
ary of degree of the vertices, i.e., 2 , for all
1 ≤ i ≤ n.
deg(bi )l +deg(b j )h
Step 5 Select maximum { 2 } for all 1 ≤ i ≤
n.

6 Application: criteria’s in education system


in Taiwan in terms of IVFG

IVFG is an important factor to know the information in many


real connected graphical system, in which the vertices and
edges both are lies in an interval. In this section, using the
help of sum of the lower and upper boundary of the vertices,
i.e., 21 [degl (a) + degu (a)] for all vertices a in an IVFG G,
we can arrange the vertices in an IVFG in Taiwan education
system for communication.

6.1 Model construction


Fig. 15 Flowchart of the proposed algorithm Taiwan’s higher education has experienced a quick devel-
opment over the past decade and that time the number of
institutions was at the university level has been reached 160.
5 Algorithm to calculate the half of the sum Facing serve competition, all this type of institutions have
of degrees of vertices in an IVFG strived to maintain one of the best position by proposing
quality teaching, services and research. Contrary to this back-
An algorithm to determine the half of sum of the upper and drop, the Taiwan Assessment and Evaluation Association
lower boundary of degree of the vertices of an IVFG is pro- (TWAEA) a profitless organization that was consecrated in
posed. Using the Dijkstra’s algorithm (Douglas 2002), first 2000 to serve third-party evaluations of the execution of sev-
draw a crisp graph G ∗ = (V , E), where V and E denote eral universities in Taiwan. The TWAEA was founded by the
the set of all vertices and edges with |V | = n and |E| = m, senior members of business sectors and academia.
respectively (Fig. 15). To assist every institution for understanding the oppor-
Step 1 Put the membership values μ A (bi ) = [μlA (bi ), tunities, strengths, threats and weakness of its subservient
μ A (bi )] ⊆ [0, 1] of the vertices bi ’s, i = 1, 2, . . . , n,
h departments, the TWAEA tried to establish various achieve-
μhA (bi ) ∈ [0, 1], μlA (bi ) ∈ [0, 1] with μhA (bi ) ≥ μlA (bi ). ment evaluation models for different academic departments.

123
Pragmatic results in Taiwan education system based IVFG & IVNG

tionship, to achieve the hierarchical evaluation framework,


was proposed by the committee. Here, we represent the total
process as an IVFG G, which is exhibited in Fig. 16. Ver-
tices of G denote the criteria’s TI, TQ, TM, JP, RG, PA,
AA, SC, PS and US. Since the committee members will be
able to evaluate and acknowledge their acceptance using only
an incompletely included interval values, so the vertices are
uncertain and the membership values of the vertices are given
in Guha et al. (2009). The membership values of vertices are
given in Table 2. The committee members were asked to
compare for similarity between these criteria’s. The edges
represent the relation between every pair of vertices, which
Fig. 16 An IVFG G whose membership values of vertices are given in are also uncertain and interval valued. The membership val-
Table 2 and membership values of edges are given in Table 3 ues of edges are given in Table 3.

Table 2 Membership values of vertices of the IVFG of Fig. 16


6.3 Decision making
TI TQ TM JP RG
(0.6, 0.9) (0.6, 0.8) (0.5, 0.7) (0.4, 0.8) (0.5, 0.9)
From Table 4, it turns out that, the upper and lower value
AA PA SC PS US
of the interval of degree of some vertices is greater than or
(0.2, 0.7) (0.4, 0.9) (0.3, 0.8) (0.5, 0.7) (0.6, 0.8)
less than or equal to the upper and lower value of the interval
of degree of the other vertices. Again, the interval of degree
of some vertices totally belongs to the interval of degree
A hierarchical evaluation framework was adopted, using of some other vertices. So from here, the order of the ver-
the engineering school as an example and considering var- tices of G cannot be determined. Thus, using the half of sum
ious aspects of departmental development. For evaluating of the upper and lower boundary of degree of the vertices,
the performance, there are 10 criteria was used: teaching we can write the vertices in ascending or descending order
innovations, teaching quality, journal paper, teaching mate- to their contributions. Therefore, from Table 5 we have the
rial, research grant, patent acquisition, academic award, descending order of the vertices according to the half of sum
student consultation, university service and professional ser- of the lower and upper boundary of degree of the vertices
vice which are denoted by TI, TQ, JP, TM, RG, PA, AA, SC, of G which are US, TQ or PA, TM, RG, TI or PS, JP, SC,
US and PS, respectively. AA. Hence, there are four types of order of the ten criteria’s
which are : (i) US, TQ, PA, TM, RG, TI, PS, JP, SC, AA; (ii)
6.2 Illustration of membership values US, TQ, PA, TM, RG, PS, TI, JP, SC, AA; (iii) US, PA, TQ,
TM, RG, PS, TI, JP, SC, AA; (iv) US, PA, TQ, TM, RG, TI,
At the beginning, the core discussions focused on this fact that PS, JP, SC, AA.
how to make a group of these criteria to set up a hierarchical Therefore, the criteria US, i.e., University Service main-
framework for evaluation. A clustering based on fuzzy rela- tained is the most significant criteria.

Table 3 Membership values of


edges of the IVFG of Fig. 16 TI-TQ TI-TM TI-JP TI-RG TI-AA TI-PA TI-SC TI-PS
(0.5, 0.6) (0.4, 0.5) (0.2, 0.4) (0.3, 0.6) (0.2, 0.7) (0.4, 0.5) (0.2, 0.6) (0.4, 0.5)
TI-US TQ-TM TQ-JP TQ-RG TQ-AA TQ-PA TQ-SC TQ-PS
(0.6, 0.7) (0.4, 0.7) (0.4, 0.6) (0.3, 0.6) (0.2, 0.6) (0.3, 0.5) (0.3, 0.7) (0.4, 0.6)
TQ-US TM-JP TM-RG TM-AA TM-PA TM-SC TM-PS TM-US
(0.3, 0.6) (0.3, 0.6) (0.4, 0.6) (0.1, 0.7) (0.4, 0.6) (0.2, 0.7) (0.3, 0.6) (0.4, 0.6)
JP-RG JP-AA JP-PA JP-SC JP-PS JP-US RG-AA RG-PA
(0.4, 0.7) (0.2, 0.7) (0.3, 0.7) (0.3, 0.5) (0.4, 0.6) (0.3, 0.6) (0.2, 0.6) (0.4, 0.8)
RG-SC RG-PS RG-US AA-PA AA-SC AA-PS AA-US PA-SC
(0.2, 0.5) (0.2, 0.5) (0.4, 0.7) (0.1, 0.7) (0.2, 0.6) (0.2, 0.7) (0.1, 0.5) (0.2, 0.6)
PA-PS PA-US SC-PS SC-US PS-US
(0.3, 0.6) (0.4, 0.8) (0.4, 0.6) (0.2, 0.7) (0.3, 0.7)

123
S. Poulik et al.

Table 4 Degree of vertices of


the IVFG of Fig. 16 deg(TI) deg(TQ) deg(TM) deg(JP) deg(RG)
= (3.2, 5.1) = (3.1, 5.5) = (2.9, 5.6) = (2.8, 5.4) = (2.8, 5.6)
deg(AA) deg(PA) deg(SC) deg(PS) deg(US)
= (1.5, 5.8) = (2.8, 5.8) = (2.2, 5.5) = (2.9, 5.4) = (3.0, 5.9)

Table 5 Half of sum of lower


deg(TI)l +deg(TI)u deg(TQ)l +deg(TQ)u deg(TM)l +deg(TM)u
and upper boundary of the 2 = 4.15 2 = 4.3 2 = 4.25
degree of vertices of the IVFG deg(JP)l +deg(JP)u deg(RG)l +deg(RG)u deg(AA)l +deg(AA)u
G 2 = 4.1 2 = 4.2 2 = 3.65
deg(PA)l +deg(PA)u deg(SC)l +deg(SC)u deg(PS)l +deg(PS)u
2 = 4.3 2 = 3.85 2 = 4.15
deg(US)l +deg(US)u
2 = 4.45

7 Hypothesis and limitations • A real application of IVFG is presented to maintain an


educational system.
The main hypothesis of this work is to found out the main
characteristics in an education system for evaluation struc- Some of the disadvantages of this study are as follows:
ture. All the characteristics are connected by a graphical
structure in terms of interval-valued fuzzy graph structure.
We pointed out that there are some defects in the definitions • All the data must be given. But collection of membership
of some operations of IVFG and in the definition of IVNG values may not be easy.
and its complement and strong IVNG. These definitions and • If the membership values are crisp or binary process, then
operations needed to be corrected. Then, the selection of the system reduces to hard computing.
important characteristics is shown using the degree of ver- • These results are applicable only to IVFG and IVNG.
tices in an IVFG.
Some of the limitations of this study are as follows:
9 Conclusion
• This work mainly focused on IVFG & IVNG. Graphical structure is one of the most cited model to describe
• Collecting all the data may not be easy in these environ- complicated problems in which relationships between all
ment. related criteria are similar and uncertain but not fixed. IVFG
• If the membership values of the character’s are given in and its operations give more precision, where the vertices
different environment, then IVFG cannot be used. and edges continue at a certain interval in fuzzy graph.
First, this manuscripts identify many defects on the union,
join of two IVFGs and complete IVFG using many exam-
8 Discussion: advantages and disadvantages ples. Improved versions of these definitions are established
for soft computing with examples. Also, the flaws in the definition of IVNG
have been pointed out using counterexample and its gen-
Soft computing is mainly used for approximations, and it eral form is initiated. Using this definition of IVNG, it is
gives imprecise and usable solution for solving complex com- determined that the definitions of complete and strong IVNG
putational problems. As a role model in human mind, soft are also not correct. Modified versions of these are also
computing is tolerant of uncertainty or fuzzy logic, approxi- given. Fourth, an algorithm and a flowchart of the proposed
mation and imprecision unlike traditional computers models. algorithm are established to evaluate the highest impor-
The main advantages are tant vertices in an IVFG system. Fifth, an application of
IVFG is depicted for decision making in the Taiwan edu-
cation system. Using the degree of vertices of the IVFG
• This work mainly depends on fuzzy logic in a network corresponding to the education system, the most valuable
system, so this work is compatible with soft computing. criteria for feature are shown. Sixth, the main hypothesis
• Many important definitions are improved in this study with limitations of this work has been described. Lastly,
which are very useful. some of the important advantages and disadvantages are dis-
• It contains algorithm, flowchart which are the other cussed.
important parts of soft computing.

123
Pragmatic results in Taiwan education system based IVFG & IVNG

Acknowledgements The first author is thankful to the Department of Bustince H, Burillo P (2000) Mathematical analysis of interval-valued
Higher Education, Science and Technology and Biotechnology, Gov- fuzzy relations: application to approximate reasoning. Fuzzy Sets
ernment of West Bengal, India, for the award of Swami Vivekananda Syst 113(2):205–219
merit-cum-means scholarship (Award No. 52-Edn (B)/5B-15/2017 Cornelis C, Deschrijver G, Kerre EE (2006) Advances and challenges
dated 07/06/2017) to meet up the financial expenditure to carry out the in interval-valued fuzzy logic. Fuzzy Sets Syst 157(5):622–627
research work. The third author is supported by Research Council Faroe Douglas BW (2002) Introduction to graph theory. Pearson Education
Islands and University of the Faroe Islands. The authors are grateful to India, Noida
the learned reviewers for their valuable comments and suggestions to Ghorai G, Pal M (2016) Some isomorphic properties of m-polar fuzzy
improve the quality of the article. graphs with applications. SpringerPlus. https://doi.org/10.1186/
s40064-016-3783-z
Ghorai G, Pal M (2017) Certain types of product bipolar fuzzy graphs.
Compliance with ethical standards Int J Appl Comput Math 3(2):605–619
Ghorai G, Pal M (2018) A note on “Regular bipolar fuzzy graphs”
Neural Computing and Applications 21(1) 2012 197–205. Neural
Conflict of interest The authors declare that they have no conflict of Comput Appl 30(5):1569–1572
interest. Guha YY, Yangb MS, Poc RW, Lee S (2009) Interval-valued fuzzy
relation-based clustering with its application to performance eval-
Ethical approval This article does not contain any studies with human uation. Comput Math Appl 57(5):841–849
participants or animals performed by any of the authors. Jan N, Ullah K, Mahmood T, Garg H, Davvaz B, Saeid AB, Broumi
S (2019) Some root level modifications in interval valued fuzzy
graphs and their generalizations including neutrosophic graphs.
Mathematics 7(1):72
Mahdi MA, Al-Janabi S (2020) A nobel software to improve health-
care base on predictive analytics and mobile services for cloud
data centers. In: Farhaoui Y (ed) Big data and networks technolo-
References gies. BDNT 2019, vol 81. Lecture notes in networks and systems.
Springer, Cham. https://doi.org/10.1007/978-3-030-23672-4_8
Akram M, Dudek WA (2011) Interval-valued fuzzy graphs. Comput Manne F, Xin Q (2008) Time effcient radio broadcasting in planar
Math Appl 61(2):289–299 graphs. J Netw 3(2):9–16
Akram M, Nasir M (2017) Concepts of interval-valued neutrosophic Mezei J, Wikstrom R (2013) Aggregation operators and interval-valued
graphs. Int J Algebra Stat 6(1–2):22–41 fuzzy numbers in decision making. Adv Inf Syst Technol 206:535–
Ali SH (2013) Novel approach for generating the key of stream cipher 544
system using random forest data mining algorithm. In: IEEE, 2013 Mordeson JN, Peng CS (1994) Operation on fuzzy graphs. Inf Sci
sixth international conference on developments in esystems engi- 79:159–170
neering, Abu Dhabi, 2013, pp 259–269. https://doi.org/10.1109/ Peng X, Dai J (2018) Approaches to single-valued neutrosophic MADM
DeSE.2013.54 based on MABAC, TOPSIS and new similarity measure with score
Al-Janabi S (2020) Smart system to create an optimal higher education function. Neural Comput Appl 29(10):939–954
environment using IDA and IOTs. Int J Comput Appl. https://doi. Poulik S, Ghorai G (2020) Certain indices of graphs under bipolar fuzzy
org/10.1080/1206212X.2018.1512460 environment with applications. Soft Comput 24(7):5119–5131
Al-Janabi S, Alkaim AF (2020) A nifty collaborative analysis to pre- Poulik S, Ghorai G (2020) Detour g-interior nodes and detour g-
dicting a novel tool (DRFLLS) for missing values estimation. Soft boundary nodes in bipolar fuzzy graph with applications. Hacet
Comput 24(1):555–569 J Math Stat 49(1):106–119
Al-Janabi S, Alkaim AF, Adel Z (2020) An innovative synthesis Poulik S, Ghorai G (2020) Note on “bipolar fuzzy graphs with applica-
of deep learning techniques (DCapsNet & DCOM) for genera- tions”. Knowl Based Syst. https://doi.org/10.1016/j.knosys.2019.
tion electrical renewable energy from wind energy. Soft Comput 105315
24:10943–10962 Rashmanlou H, Jouybari MN (2017) New concepts of interval-valued
Al-Janabi S, Mohammad M, Al-Sultan A (2020) A new method for fuzzy graphs with application. J Math Inform 8:69–81
prediction of air pollution based on intelligent computation. Soft Rosenfield A (1975) Fuzzy graphs. In: Zadeh LA, Fu KS, Shimura M
Comput 24:661–680 (eds) Fuzzy sets and their application. Academic press, New York,
Al-Janabi S, Yaqoob A, Mohammad M (2020) Pragmatic method based pp 77–95
on intelligent big data analytics to prediction air pollution. In: Talebi AA, Rashmanlou H, Ameri R (2016) New concepts of product
Farhaoui Y (ed) Big data and networks technologies. BDNT 2019, interval-valued fuzzy graph. J Appl Math Inform 34(3–4):179–192
vol 81. Lecture notes in networks and systems. Springer, Cham. Wang H, Smarandache F, Zhang YQ, Smarandache R (2005) Interval
https://doi.org/10.1007/978-3-030-23672-4_8 neutrosophic sets and logic: theory and applications in computing.
Alkaim AF, Al-Janabi S (2020) Multi objectives optimization to gas Hexis, Phoenix
flaring reduction from oil production. In: Farhaoui Y (ed) Big data Wang H, Smarandache F, Zhang YQ, Sunderraman R (2012) Single
and networks technologies. BDNT 2019, vol 81. Lecture notes in valued neutrosophic sets. Tech Sci Appl Math 4:410–413
networks and systems. Springer, Cham. https://doi.org/10.1007/ Xin Q, Xia Y (2017) Latency optimal broadcasting in noisy wireless
978-3-030-23672-4_10 mesh networks. In: 2017 27th International telecommunication
Bhutani KR, Rosenfeld A (2003) Strong arc in fuzzy graphs. Inf Sci networks and applications conference (ITNAC). https://doi.org/
152:319–322 10.1109/ATNAC.2017.8215431
Broumi S, Deli I, Smarandache F (2014) Relations on interval valued Yang HL, Guo ZL, She Y, Liao X (2016) On single valued neutrosophic
neutrosophic soft sets. J New Results Sci. https://doi.org/10.5281/ relations. J Intell Fuzzy Syst 30(2):1045–1056
zenodo.30306 Yang HL, Zhang CL, Guo ZL, Liu YL, Liao X (2017) A hybrid model
Broumi S, Talea M, Bakali A, Smarandache F (2016) Single valued of single valued neutrosophic sets and rough sets: single valued
neutrosophic graphs. J New Theory 10:86–101 neutrosophic rough set model. Soft Comput 21(21):6253–6267

123
S. Poulik et al.

Yu CW, Xin Q, Chilamkurti N, Jiang S (2014) Algorithm and theory for


robust wireless sensor networks. Int J Distrib Sens Netw 10(4):1–1
Zadeh LA (1965) Fuzzy sets, information. Control 8(3):338–353
Zhang H, Wang J, Chen X (2014) Interval neutrosophic sets and their
application in multicriteria decision making problems. Hindawi
Publishing Corporation, London. https://doi.org/10.1155/2014/
645953

Publisher’s Note Springer Nature remains neutral with regard to juris-


dictional claims in published maps and institutional affiliations.

123

You might also like