International Journal of Computational and Applied Mathematics

ISSN 1819-4966 Volume 4 Number 2 (2009), pp. 95–106

© Research India Publications
http://www.ripublication.com/ijcam.htm

Comparative Study of Rough Sets on Fuzzy

Approximation Spaces and Intuitionistic Fuzzy
Approximation Spaces

D. P. Acharjya

School of Computing Sciences, VIT University, Vellore 632014,

Tamilnadu, INDIA
E-mail: [email protected]

Abstract

The concept of fuzzy approximation space that depends on a fuzzy proximity

relation is a generalization of the concept of the knowledge base. But
intuitionistic fuzzy approximation space that depends on an intuitionistic fuzzy
proximity relation is a better generalization of the concept of knowledge base
than fuzzy approximation space. Therefore, rough sets defined on intuitionistic
fuzzy approximation spaces extend the concept of rough sets on fuzzy
approximation spaces. This paper presents how rough sets on intuitionistic
fuzzy approximation spaces provides better result over rough sets on fuzzy
approximation spaces on knowledge representation.

Index Terms: Fuzzy relation, fuzzy proximity relation, fuzzy approximation

space, intuitionistic fuzzy approximation space and rough set.

1. Introduction
The concept of classical sets has been extended in many directions as far as modelling
of real life situations are concerned. L. A. Zadeh [12] extended the concept of
classical set to the notion of fuzzy sets that captures impreciseness in information. On
the other hand on introducing rough sets Z. Pawlak [6] captured indeiscernibility
among objects to model imperfect knowledge [7], [8], [9]. A set of data is generally
disorganized, especially at the age of Internet. On the other end much useful
knowledge is hidden in the accumulated voluminous data. Therefore, changing data
into knowledge is not a straight forward task. Rough set, which is based upon the
approximation of a set by a pair of sets with respect to some imprecise information, is
a formal framework for the automated transformation of data into knowledge.
96 D. P. Acharjya

Rough sets, as introduced by Z. Pawlak depend upon the concept of equivalence

relations to classify objects. However in many real life situations, pure classification
is rarely possible because of the transitivity being involved. So, rough sets on fuzzy
approximation spaces is defined by De [4] by taking a fuzzy approximation space
which depends upon the concept of fuzzy proximity relation. In fuzzy sets, the
membership and non-membership of elements are supposed to be one’s complement
of each other, but in many real life cases it is not true. As a result Atanasov [3]
introduced the concept of intuitionistic fuzzy sets. This further studied in a different
dimension by Tripathy [10] on intuitionistic fuzzy approximation spaces (IF-
approximation space) which depend upon the concept of intuitionistic fuzzy proximity
relation (IF-proximity relation). Then rough sets on IF-approximation spaces are
studied for better classifications of objects.
Types of rough sets is a topological property, introduced by Pawlak [6]. The types
of rough sets obtained as union or intersection of rough sets was studied in detail by
Tripathy and Mitra [11], which was extended to the concept of rough sets on fuzzy
approximation spaces by Acharjya and Tripathy [2]. Again it is further extended to
the setting of rough sets on intuitionistic fuzzy approximation spaces by Acharjya and
Tripathy [1].
In this paper, we provide a real life example of ranking different institutes
depending upon some characteristics and compare how rough sets on intuitionistic
fuzzy approximation spaces gives better classification over rough sets on fuzzy
approximation spaces.

2. Rough Sets on Fuzzy Approximation Space

We unfold the background of this article in this section, by presenting the
fundamental concepts, notations and results on rough sets on fuzzy approximation
spaces, and these are all bases our discussion starts from.

Definition 2.1 Let U be a universe. We define a fuzzy relation on U as a fuzzy subset

of (U × U).

Definition 2.2 A fuzzy relation R on U is said to be a fuzzy proximity relation if

μ R ( x, x) = 1 for all x ∈ U (2.1)
and μ R ( x, y ) = μ R ( y, x) for x, y ∈ U. (2.2)

Definition 2.3 Let R be a fuzzy proximity relation on U. We say that two elements x
and y are α -similar, α ∈ [0, 1], with respect to R if ( x, y ) ∈ Rα and we write xRα y .

Definition 2.4 Two elements x and y are said to be α -identical with respect to R if
either x is α -similar to y or x is transitively α -similar to y with respect to R, i.e., there
exists a sequence u1 , u2 , u3 ,L , un in U such that xRα u1 , u1Rα u2 , u2 Rα u3 , L , un Rα y . If
Comparative Study of Rough Sets on Fuzzy 97

x and y are α -identical with respect to R, then we write xR (α ) y , where the relation
R (α ) , for each fixed α ∈ [0, 1] is an equivalence relation on U.

Definition 2.5 The pair (U, R) is called a fuzzy approximation space. For any α ∈ [0,
1], we denote by Rα* , the set of all equivalence classes of R (α ) . Also, we call
(U, R (α )) , the generated approximation space associated with R and α.

Definition 2.6 Let (U, R) be a fuzzy approximation space and let X ⊆ U . Then the
rough set of X in (U, R (α )) is denoted by ( X α , X α ) , where X α is the α-lower
approximation of X whereas X α is the α-upper approximation of X. We define X α
and X α as
X α = U{Y : Y ∈ Rα* and Y ⊆ X } (2.3)
and X α = U{Y : Y ∈ Rα* and Y I X ≠ φ} (2.4)

Definition 2.7 Let X ⊆ U . Then X is said to be α-discernible if and only if X α = X α

and X is said to be α-rough if and only if X α ≠ X α .
Many properties of α-lower and α-upper approximations have been studied by De
[4]. In the next section, we shall be extending these results in the context of IF-
approximation space.

3. Rough Sets on If-Approximation Space

Atanasov [3] introduced intuitionistic fuzzy sets as an extension of fuzzy sets. Fuzzy
relations and intuitionistic fuzzy relations are extension of the concept of crisp
relations. Keeping the background of this paper, we present the fundamental concepts,
notations and results on rough sets on IF-Approximation Space as developed by
Tripathy [10]. We first define the following concepts leading to the introduction of
rough sets on IF-approximation space. We use standard notation μ for membership
and ν for non-membership functions associated with an intuitionistic fuzzy set.

Definition 3.1 An intuitionistic fuzzy relation on U is an intuitionistic fuzzy subset of

(U × U).

Definition 3.2 An intuitionistic fuzzy relation R on U is said to be an intuitionistic

fuzzy proximity relation if
μ R ( x, x) = 1 and ν R ( x, x) = 0 ∀x ∈ U (3.1)
and μ R ( x, y ) = μ R ( y, x),ν R ( x, y ) = ν R ( y, x) ∀x, y ∈ U (3.2)
We write J = {(α , β ) α , β ∈ [0, 1] and 0 ≤ α + β ≤ 1}
98 D. P. Acharjya

Definition 3.3 Let R be an IF-proximity relation on U. Then for any (α , β ) ∈ J the

(α , β ) -cut ' Rα , β ' of R is given by
Rα , β = {( x, y ) μ R ( x, y ) ≥ α and ν R ( x, y ) ≤ β } (3.3)

Definition 3.4 Let R is an IF-proximity relation on U. We say that two elements x and
y are (α , β ) - similar with respect to R if ( x, y ) ∈ Rα , β and write x Rα , β y .

Definition 3.5 Let R is an IF-proximity relation on U. We say that two elements x and
y are (α , β ) - identical with respect to R for (α , β ) ∈ J , written as xR (α , β ) y if and
only if x Rα , β y or there exists a sequence of elements u1 , u2 , u3 ,L , un in U such that
x Rα , β u1 , u1 Rα , β u2 , u2 Rα , β u3 ,L , un Rα , β y . In this case we say that x is transitively
(α , β ) -similar to y with respect to R.

Note 3.1 It is easy to see that for any (α , β ) ∈ J , R (α , β ) is an equivalence relation

on U.
We use Rα* , β to denote the set of equivalence classes generated by the equivalence
relation R (α , β ) for any (α , β ) ∈ J . We denote the R (α , β ) -equivalence class of an
element x in U by [ x]α , β . The pair (U, R) is called an intuitionistic fuzzy
approximation space (IF-approximation space)

Note 3.2 An IF-approximation space (U, R) generates usual approximation space

(U, R (α , β )) of Pawlak for every (α , β ) ∈ J .

Definition 3.6 The rough set on X in the generalized approximation space

(U, R (α , β )) is denoted by ( X α , β , X α , β ) where
X α , β = U{Y Y ∈ Rα* , β and Y ⊆ X } (3.4)
and X α , β = U{Y Y ∈ Rα* , β and Y ∩ X ≠ φ} (3.5)

Definition 3.7 Let X be a rough set in the generalized approximation space

(U , R (α , β )) . Then we define the (α , β ) - boundary of X with respect to R i.e.,
BNRα , β ( X ) as
BNRα , β ( X ) = X α , β − X α , β (3.6)

Definition 3.8 Let X be a rough set in the generalized approximation space

(U , R (α , β )) . Then we say that
X is (α , β ) -discernible with respect to R if and only if X α , β = X α , β (3.7)
and
Comparative Study of Rough Sets on Fuzzy 99

X is (α , β ) -rough with respect to R if and only if X α , β ≠ X α , β . (3.8)

The following theorems of Tripathy [10] establish properties of (α , β ) -lower
approximation and (α , β ) -upper approximation sets of rough set.
Let X and Y be rough sets in the generalized approximation space (U, R(α , β )) ,
(α , β ) ∈ J . Then
X α ,β ⊆ X ⊆ X α ,β (3.9)
φ α ,β = φ = φ α ,β , U α ,β = U = U α ,β (3.10)
( X ∪ Y )α , β = X α , β ∪ Y α , β (3.11)
( X ∩ Y )α , β = X α , β ∩ Y α , β (3.12)
X ⊆ Y ⇒ X α ,β ⊆ Y α ,β (3.13)
X ⊆ Y ⇒ X α ,β ⊆ Y α ,β (3.14)
( X ∪ Y )α , β ⊇ X α , β ∪ Y α , β (3.15)
( X ∩ Y )α , β ⊆ X α , β ∩ Y α , β (3.16)
If α1 ≥ α 2 and β1 ≤ β 2 , then
X α 2 , β2 ⊆ X α1 , β1 (3.17)
and X α1 , β1 ⊆ X α 2 , β2 (3.18)

4. Comparative Study On Knowledge Representations

In this section, we demonstrate how rough sets on intuitionistic fuzzy approximation
spaces provide better representation of knowledge than rough sets on fuzzy
approximation spaces. We consider the same example of ranking of institutions in a
country introduced in [1]. However, for completeness we shall be providing the
details of the example below. In this example five parameters were considered which
control the performance of institutions and are taken as their measures to rank them.
In the table below we specify parameters, their possible range of values and an
intuitionistic fuzzy proximity relation which characterizes the relationship among
parameters.
The institute with more intellectual capital, having high infrastructure, better
placement opportunities for students in high profile organizations and higher student
satisfaction measure is an ideal case for highest rank. But it may not be possible for an
institute to excel in all the above fields. However, some of the parameters considered
above may have higher influence on the rank than other parameters. The importance
of parameters depends upon the control parameters and so varies along with the
values of these parameters. In fact, it has been observed [1] that if we increase the
values of α and β more and more parameters becomes indispensable.
The membership and non-membership functions have been adjusted such that the
sum of their values should lie in [0, 1] and also these functions must be symmetric.
100 D. P. Acharjya

The first requirement necessitates a major of 2 in the denominators of the non-

membership functions.
The institutes can be judged by the outputs, which are produced. The quality of
the output can be judged by the placement performance of the institute and is given
highest weight with a score 385 which comes to around 24% of total weight. To
produce the quality output the input should be of high quality and the major inputs for
an institute are intellectual capital and infrastructure facilities required for imparting
education. Accordingly the scores for intellectual capital and an infrastructure facility
are fixed as 250 and 200 respectively that weight 15% and 12% of total weight. The
student placed in the company shall serve the company up to their expectation, i.e.,
the student shall meet the recruiter’s satisfaction and is given with a score of 200
which comes to around 12%. At the same time students satisfaction is given with a
score 60 of weight 4%. However we have not considered many other parameters that
do not influence the ranking of the institutions and to make our analysis simple.

Parameter Attribute Possible Membership Non- membership

Range Function Function
Intellectual IC 1 – 250 x−y x− y
1− 250 2( x + y )
Capital
Infrastructure IF 1 – 200 x−y x− y
1− 200 2( x + y )
Facility
Placement PP 1 – 385 x−y x− y
1− 385 2( x + y )
Performance
Recruiters RS 1 – 200 x−y x− y
1− 200 2( x + y )
Satisfaction
Score
Students SS 1 – 60 x−y x− y
1− 60 2( x + y )
Satisfaction
Score

We consider a small universe of 10 institutions and the information pertaining to them

are presented in the following table.

Institution IC IF PP RS SS
I1 229 151 303 169 56
I2 226 145 266 167 54
I3 227 142 298 169 53
I4 179 117 247 160 53
I5 188 135 251 152 51
I6 130 120 209 149 48
I7 100 113 151 143 39
I8 97 115 245 146 47
I9 96 75 144 147 40
I10 110 59 138 139 42
Comparative Study of Rough Sets on Fuzzy 101

IF-proximity relation R1 corresponding to attribute ‘IC’ is given as:

R1 I1 I2 I3 I4 I5 I6 I7 I8 I9 I10
I1 1.000, 0.988, 0.992, 0.800, 0.836, 0.604, 0.484, 0.472, 0.468, 0.524,
0.000 0.003 0.002 0.061 0.049 0.138 0.196 0.202 0.205 0.176
I2 0.988, 1.000, 0.996, 0.812, 0.848, 0.616, 0.496, 0.484, 0.480, 0.536,
0.003 0.000 0.001 0.058 0.045 0.135 0.193 0.200 0.202 0.173
I3 0.992, 0.996, 1.000, 0.808, 0.844, 0.612, 0.492, 0.480, 0.476, 0.532,
0.002 0.001 0.000 0.059 0.047 0.136 0.194 0.201 0.203 0.174
I4 0.800, 0.812, 0.808, 1.000, 0.964, 0.804, 0.684, 0.672, 0.668, 0.724,
0.061 0.058 0.059 0.000 0.012 0.079 0.142 0.149 0.151 0.119
I5 0.836, 0.848, 0.844, 0.964, 1.000, 0.768, 0.648, 0.636, 0.632, 0.688,
0.049 0.045 0.047 0.012 0.000 0.091 0.153 0.160 0.162 0.131
I6 0.604, 0.616, 0.612, 0.804, 0.768, 1.000, 0.880, 0.868, 0.864, 0.920,
0.138 0.135 0.136 0.079 0.091 0.000 0.065 0.072 0.075 0.041
I7 0.484, 0.496, 0.492, 0.684, 0.648, 0.880, 1.000, 0.988, 0.984, 0.960,
0.196 0.193 0.194 0.142 0.153 0.065 0.000 0.007 0.010 0.023
I8 0.472, 0.484, 0.480, 0.672, 0.636, 0.868, 0.988, 1.000, 0.996, 0.948,
0.202 0.200 0.201 0.149 0.160 0.072 0.007 0.000 0.002 0.031
I9 0.468, 0.480, 0.476, 0.668, 0.632, 0.864, 0.984, 0.996, 1.000, 0.944,
0.205 0.202 0.203 0.151 0.162 0.075 0.010 0.002 0.000 0.034
I10 0.524, 0.536, 0.532, 0.724, 0.688, 0.920, 0.960, 0.948, 0.944, 1.000,
0.176 0.173 0.174 0.119 0.131 0.041 0.023 0.031 0.034 0.000

IF-proximity relation R 2 corresponding to attribute ‘IF’ is given as:

R2 I1 I2 I3 I4 I5 I6 I7 I8 I9 I10
I1 1.000, 0.970, 0.955, 0.830, 0.920, 0.845, 0.810, 0.820, 0.620, 0.540,
0.000 0.010 0.015 0.063 0.028 0.057 0.072 0.067 0.168 0.219
I2 0.970, 1.000, 0.985, 0.860, 0.950, 0.875, 0.840, 0.850, 0.650, 0.570,
0.010 0.000 0.005 0.053 0.017 0.047 0.062 0.057 0.159 0.211
I3 0.955, 0.985, 1.000, 0.875, 0.965, 0.890, 0.855, 0.865, 0.665, 0.585,
0.015 0.005 0.000 0.048 0.012 0.042 0.056 0.052 0.154 0.206
I4 0.830, 0.860, 0.875, 1.000, 0.910, 0.985, 0.980, 0.990, 0.790, 0.710,
0.063 0.053 0.048 0.000 0.035 0.006 0.008 0.004 0.109 0.165
I5 0.920, 0.950, 0.965, 0.910, 1.000, 0.925, 0.890, 0.900, 0.700, 0.620,
0.028 0.017 0.012 0.035 0.000 0.029 0.044 0.040 0.143 0.196
I6 0.845, 0.875, 0.890, 0.985, 0.925, 1.000, 0.965, 0.975, 0.775, 0.695,
0.057 0.047 0.042 0.006 0.029 0.000 0.015 0.010 0.115 0.170
I7 0.810, 0.840, 0.855, 0.980, 0.890, 0.965, 1.000, 0.990, 0.810, 0.730,
0.072 0.062 0.056 0.008 0.044 0.015 0.000 0.004 0.101 0.157
I8 0.820, 0.850, 0.865, 0.990, 0.900, 0.975, 0.990, 1.000, 0.800, 0.720,
0.067 0.057 0.052 0.004 0.040 0.010 0.004 0.000 0.105 0.161
I9 0.620, 0.650, 0.665, 0.790, 0.700, 0.775, 0.810, 0.800, 1.000, 0.920,
0.168 0.159 0.154 0.109 0.143 0.115 0.101 0.105 0.000 0.059
102 D. P. Acharjya

I10 0.540, 0.570, 0.585, 0.710, 0.620, 0.695, 0.730, 0.720, 0.920, 1.000,
0.219 0.211 0.206 0.165 0.196 0.170 0.157 0.161 0.059 0.000

IF-proximity relation R 3 corresponding to attribute ‘PP’ is given as:

R3 I1 I2 I3 I4 I5 I6 I7 I8 I9 I10
I1 1.000, 0.904, 0.987, 0.855, 0.865, 0.756, 0.605, 0.849, 0.587, 0.571,
0.000 0.032 0.004 0.050 0.046 0.091 0.167 0.052 0.178 0.187
I2 0.904, 1.000, 0.917, 0.951, 0.961, 0.852, 0.701, 0.945, 0.683, 0.668,
0.032 0.000 0.028 0.018 0.014 0.060 0.138 0.020 0.149 0.158
I3 0.987, 0.917, 1.000, 0.868, 0.878, 0.769, 0.618, 0.862, 0.600, 0.584,
0.004 0.028 0.000 0.046 0.042 0.087 0.164 0.048 0.174 0.183
I4 0.855, 0.951, 0.868, 1.000, 0.990, 0.901, 0.751, 0.995, 0.732, 0.717,
0.050 0.018 0.046 0.000 0.004 0.041 0.121 0.002 0.132 0.142
I5 0.865, 0.961, 0.878, 0.990, 1.000, 0.891, 0.740, 0.984, 0.722, 0.706,
0.046 0.014 0.042 0.004 0.000 0.045 0.124 0.006 0.135 0.145
I6 0.756, 0.852, 0.769, 0.901, 0.891, 1.000, 0.849, 0.906, 0.831, 0.816,
0.091 0.060 0.087 0.041 0.045 0.000 0.080 0.039 0.092 0.102
I7 0.605, 0.701, 0.618, 0.751, 0.740, 0.849, 1.000, 0.756, 0.982, 0.966,
0.167 0.138 0.164 0.121 0.124 0.080 0.000 0.119 0.011 0.022
I8 0.849, 0.945, 0.862, 0.995, 0.984, 0.906, 0.756, 1.000, 0.738, 0.722,
0.052 0.020 0.048 0.002 0.006 0.039 0.119 0.000 0.130 0.140
I9 0.587, 0.683, 0.600, 0.732, 0.722, 0.831, 0.982, 0.738, 1.000, 0.984,
0.178 0.149 0.174 0.132 0.135 0.092 0.011 0.130 0.000 0.010
I10 0.571, 0.668, 0.584, 0.717, 0.706, 0.816, 0.966, 0.722, 0.984, 1.000,
0.187 0.158 0.183 0.142 0.145 0.102 0.022 0.140 0.010 0.000

IF-proximity relation R 4 corresponding to attribute ‘RS’ is given as:

R4 I1 I2 I3 I4 I5 I6 I7 I8 I9 I10
I1 1.000, 0.990, 1.000, 0.955, 0.915, 0.900, 0.870, 0.885, 0.890, 0.850,
0.000 0.002 0.000 0.013 0.026 0.031 0.041 0.036 0.034 0.048
I2 0.990, 1.000, 0.990, 0.965, 0.925, 0.910, 0.880, 0.895, 0.900, 0.860,
0.002 0.000 0.002 0.010 0.023 0.028 0.038 0.033 0.031 0.045
I3 1.000, 0.990, 1.000, 0.955, 0.915, 0.900, 0.870, 0.885, 0.890, 0.850,
0.000 0.002 0.000 0.013 0.026 0.031 0.041 0.036 0.034 0.048
I4 0.955, 0.965, 0.955, 1.000, 0.960, 0.945, 0.915, 0.930, 0.935, 0.895,
0.013 0.010 0.013 0.000 0.012 0.017 0.028 0.022 0.021 0.035
I5 0.915, 0.925, 0.915, 0.960, 1.000, 0.985, 0.955, 0.970, 0.975, 0.935,
0.026 0.023 0.026 0.012 0.000 0.004 0.015 0.010 0.008 0.022
I6 0.900, 0.910, 0.900, 0.945, 0.985, 1.000, 0.970, 0.985, 0.990, 0.950,
0.031 0.028 0.031 0.017 0.004 0.000 0.010 0.005 0.003 0.017
I7 0.870, 0.880, 0.870, 0.915, 0.955, 0.970, 1.000, 0.985, 0.980, 0.980,
0.041 0.038 0.041 0.028 0.015 0.010 0.000 0.005 0.006 0.007
I8 0.885, 0.895, 0.885, 0.930, 0.970, 0.985, 0.985, 1.000, 0.995, 0.965,
Comparative Study of Rough Sets on Fuzzy 103

0.036 0.033 0.036 0.022 0.010 0.005 0.005 0.000 0.001 0.012
I9 0.890, 0.900, 0.890, 0.935, 0.975, 0.990, 0.980, 0.995, 1.000, 0.960,
0.034 0.031 0.034 0.021 0.008 0.003 0.006 0.001 0.000 0.014
I10 0.850, 0.860, 0.850, 0.895, 0.935, 0.950, 0.980, 0.965, 0.960, 1.000,
0.048 0.045 0.048 0.035 0.022 0.017 0.007 0.012 0.014 0.000

IF-proximity relation R 5 corresponding to attribute ‘SS’ is given as:

R5 I1 I2 I3 I4 I5 I6 I7 I8 I9 I10
I1 1.000, 0.967, 0.950, 0.950, 0.917, 0.867, 0.717, 0.850, 0.733, 0.767,
0.000 0.009 0.013 0.013 0.023 0.038 0.089 0.043 0.083 0.071
I2 0.967, 1.000, 0.983, 0.983, 0.950, 0.900, 0.750, 0.883, 0.767, 0.800,
0.009 0.000 0.004 0.004 0.014 0.029 0.080 0.034 0.074 0.062
I3 0.950, 0.983, 1.000, 1.000, 0.967, 0.917, 0.767, 0.900, 0.783, 0.817,
0.013 0.004 0.000 0.000 0.009 0.024 0.076 0.030 0.069 0.057
I4 0.950, 0.983, 1.000, 1.000, 0.967, 0.917, 0.767, 0.900, 0.783, 0.817,
0.013 0.004 0.000 0.000 0.009 0.024 0.076 0.030 0.069 0.057
I5 0.917, 0.950, 0.967, 0.967, 1.000, 0.950, 0.800, 0.933, 0.817, 0.850,
0.023 0.014 0.009 0.009 0.000 0.015 0.066 0.020 0.060 0.048
I6 0.867, 0.900, 0.917, 0.917, 0.950, 1.000, 0.850, 0.983, 0.867, 0.900,
0.038 0.029 0.024 0.024 0.015 0.000 0.051 0.005 0.045 0.033
I7 0.717, 0.750, 0.767, 0.767, 0.800, 0.850, 1.000, 0.867, 0.983, 0.950,
0.089 0.080 0.076 0.076 0.066 0.051 0.000 0.046 0.006 0.018
I8 0.850, 0.883, 0.900, 0.900, 0.933, 0.983, 0.867, 1.000, 0.883, 0.917,
0.043 0.034 0.030 0.030 0.020 0.005 0.046 0.000 0.040 0.028
I9 0.733, 0.767, 0.783, 0.783, 0.817, 0.867, 0.983, 0.883, 1.000, 0.967,
0.083 0.074 0.069 0.069 0.060 0.045 0.006 0.040 0.000 0.012
I10 0.767, 0.800, 0.817, 0.817, 0.850, 0.900, 0.950, 0.917, 0.967, 1.000,
0.071 0.062 0.057 0.057 0.048 0.033 0.018 0.028 0.012 0.000

IF-proximity relation R corresponding to all attributes is given as:

R I1 I2 I3 I4 I5 I6 I7 I8 I9 I10
I1 1.000, 0.904, 0.950, 0.800, 0.836, 0.604, 0.484, 0.472, 0.468, 0.524,
0.000 0.032 0.015 0.063 0.049 0.138 0.196 0.202 0.205 0.219
I2 0.904, 1.000, 0.917, 0.812, 0.848, 0.616, 0.496, 0.484, 0.480, 0.536,
0.032 0.000 0.028 0.058 0.045 0.135 0.193 0.200 0.202 0.211
I3 0.950, 0.917, 1.000, 0.808, 0.844, 0.612, 0.492, 0.480, 0.476, 0.532,
0.015 0.028 0.000 0.059 0.047 0.136 0.194 0.201 0.203 0.206
I4 0.800, 0.812, 0.808, 1.000, 0.910, 0.804, 0.684, 0.672, 0.668, 0.710,
0.063 0.058 0.059 0.000 0.035 0.079 0.142 0.149 0.151 0.165
I5 0.836, 0.848, 0.844, 0.910, 1.000, 0.768, 0.648, 0.636, 0.632, 0.620,
0.049 0.045 0.047 0.035 0.000 0.091 0.153 0.160 0.162 0.196
I6 0.604, 0.616, 0.612, 0.804, 0.768, 1.000, 0.849, 0.868, 0.775, 0.695,
0.138 0.135 0.136 0.079 0.091 0.000 0.080 0.072 0.115 0.170
I7 0.484, 0.496, 0.492, 0.684, 0.648, 0.849, 1.000, 0.756, 0.810, 0.730,
104 D. P. Acharjya

0.196 0.193 0.194 0.142 0.153 0.080 0.000 0.119 0.101 0.157
I8 0.472, 0.484, 0.480, 0.672, 0.636, 0.868, 0.756, 1.000, 0.738, 0.720,
0.202 0.200 0.201 0.149 0.160 0.072 0.119 0.000 0.130 0.161
I9 0.468, 0.480, 0.476, 0.668, 0.632, 0.775, 0.810, 0.738, 1.000, 0.920,
0.205 0.202 0.203 0.151 0.162 0.115 0.101 0.130 0.000 0.059
I10 0.524, 0.536, 0.532, 0.710, 0.620, 0.695, 0.730, 0.720, 0.920, 1.000,
0.219 0.211 0.206 0.165 0.196 0.170 0.157 0.161 0.059 0.000

Analysis
If we give importance to all the parameters, that is, more intellectual capital having
high infrastructure, excellent placement with high recruiter and student satisfaction
score to rank an institution in a group of institutions up to particular levels of
membership and non-membership than it has to consider the common cases by taking
the intersection of all the above relations. The IF-approximation space (U, R (α , β ))
generated from all IF-proximity relations R1 , R 2 , R 3 , R 4 and R 5 corresponding to all
attributes is given above.
The equivalence classes of different relations according to intuitionistic fuzzy
approximation spaces corresponding to the value α ≥ 0.8 and β ≤ 0.1 are given by,
U R1 (α , β ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }}
U R 2 (α , β ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 },{ Ι9 , Ι10 }}
U R 3 (α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι 9 , Ι10 }}
U R 4 (α , β ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }}
U R 5 (α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι 9 , Ι10 }}
U R(α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 },{Ι9 , Ι10 }}
Now,
U (R − {R 1})(α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 },{Ι 9 , Ι10 }} = U R(α , β )
U (R − {R 2 })(α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 } ≠ U R(α , β )
U (R − {R 3 })(α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 },{Ι 9 , Ι10 }} = U R(α , β )
U (R − {R 4 })(α , β ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 },{Ι9 , Ι10 }} = U R(α , β )
U (R − {R 5 })(α , β ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 },{Ι 9 , Ι10 }} = U R(α , β )
So, it is clear from the above analysis that the relation R 2 is (α , β ) -
indispensable in R, since U (R − {R 2 })(α , β ) ≠ U R(α , β ) whereas the relation R1 is
(α , β ) - dispensable in R, since U (R − {R 1})(α , β ) = U R(α , β ) . Similarly the
relations R 3 , R 4 and R 5 are also (α , β ) - dispensable in R. That means that the
attribute infrastructure facility is significant for ranking of the institutes whereas the
attributes intellectual capital, placement performance, recruiter’s satisfaction score
and student’s satisfaction score do not play significant role in ranking of the institutes
in accordance with the control parameter values α ≥ 0.8 and β ≤ 0.1 .
Comparative Study of Rough Sets on Fuzzy 105

On the other hand if we provide importance to all parameters up to particular

levels of membership only than the equivalence classes of different relations
according to fuzzy approximation spaces corresponding to the value α ≥ 0.8 are
given by,
U R1 (α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }}
U R 2 (α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }}
U R 3 (α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι 9 , Ι10 }}
U R 4 (α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }}
U R 5 (α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι 9 , Ι10 }}
U R(α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }}
Now,
U (R − {R 1})(α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }} = U R(α )
U (R − {R 2 })(α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }} = U R(α )
U (R − {R 3 })(α ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι 9 , Ι10 }} = U R(α )
U (R − {R 4 })(α ) = {{Ι1 , Ι 2 , Ι3 , Ι 4 , Ι 5 , Ι 6 , Ι 7 , Ι8 , Ι9 , Ι10 }} = U R(α )
U (R − {R 5 })(α ) = {{Ι1 , Ι 2 , Ι 3 , Ι 4 , Ι5 , Ι 6 , Ι 7 , Ι8 , Ι 9 , Ι10 }} = U R(α )
Therefore, it is clear from the above analysis that the relations R1 , R 2 , R 3 , R 4 and
R 5 are α - dispensable in R, since U (R − {R i })(α ) = U R(α ) for i = 1, 2, 3, 4 and 5.
This means that the attributes intellectual capital, infrastructure facility, placement
performance, recruiter’s satisfaction score and student’s satisfaction score do not play
significant role in ranking of the institutes in accordance with the control parameter
value α ≥ 0.8 .
Using further analysis from rough set theory actual ranking of the institutes can
also be carried out.

5. Conclusion
We continue our study in [2] and [1] further in this paper by showing that rough sets
on intuitionistic fuzzy approximation spaces provide a better model than rough sets on
fuzzy approximation spaces for suitable values of the control parameters. To illustrate
this we continued with the example of ranking of institutes introduced and studied in
[1].

6. Acknowledgment
I express my deep sense of gratitude and indebtedness to my esteemed teacher and
supervisor Prof. B. K. Tripathy, School of Computing Sciences, VIT University for
his ingenious guidance, sincere advice and valuable suggestions in preparing this
paper. I am very thankful for the support of VIT University where I carried out this
work.
106 D. P. Acharjya

References
[1] Acharjya, D. P., and Tripathy, B. K., 2009, “Rough Sets on Intuitionistic
Fuzzy Approximation Spaces and Knowledge Representation,” International
Journal of Artificial Intelligence and Computational Research., 1(1), pp. 29-36.
[2] Acharjya, D. P., and Tripathy, B. K., 2008, “Rough Sets on Fuzzy
Approximation Spaces and Application to Distributed Knowledge Systems,”
International Journal of Artificial Intelligence and Soft Computing., 1(1), pp.
1-14.
[3] Atanasov, K. T., 1986, “Intuitionistic Fuzzy Sets,” Fuzzy Sets and Systems.,
20, pp. 87-96.
[4] De, S. K., 1999, “Some Aspects of Fuzzy sets, Rough sets and Intuitionistic
Fuzzy sets,” Ph.D. thesis, IIT, Kharagpur, India.
[5] Pawlak, Z., 1982, “Rough sets,” International Journal Comp and Information
Science., 2, pp. 341–356.
[6] Pawlak, Z., 1991, “Rough sets, Theoretical aspects of reasoning about data,”
Kluwer Academic Publishers, Dordrecht, The Netherlands.
[7] Pawlak, Z., and Skowron, A., 2007, “Rudiments of rough sets,” Information
Sciences., 177(1), pp. 3-27.
[8] Pawlak, Z., and Skowron, A., 2007, “Rough sets: Some extensions,”
Information Sciences., 177(1), pp. 28-40.
[9] Pawlak, Z., and Skowron, A., 2007, “Rough sets and Boolean reasoning,”
Information Sciences., 177(1), 41–73.
[10] Tripathy, B. K., 2006, “Rough Sets on Intuitionistic Fuzzy Approximation
Spaces,” IEEE Conference on Intelligent Systems., pp.776–779.
[11] Tripathy, B. K., and Mitra, A., “Topological Properties of Rough Sets and
Applications,” International Journal of Granular Computing, Rough Sets and
Intelligent System., communicated.
[12] Zadeh, L. A., 1965, “Fuzzy sets,” Information and Control., 8, pp. 338-353.

Biographical Notes
D P Acharjya is presently working as Associate Professor in the School of Computing
Sciences, VIT University, at Vellore, India. His 15 years of experience in higher
education include as a faculty member, reviewer, academic councilor, academic guide
and member of board of studies. He obtained his M. Tech. (Comp. Sc.) from Utkal
University, M. Phil. from Berhampur University and M. Sc. from National Institute of
Technology, Rourkela. He has been awarded with Gold Medal in M. Sc. He is a life
member of CSI, ISTE, IMS, AMTI, ISIAM, IACSIT, IAENG and Orissa Information
Technology Society (OITS). He was founder secretary of OITS Rourkela Chapter. He
has authored many national and international papers and three books to his credit. His
present area of research includes rough sets, fuzzy sets, knowledge representation,
granular computing, data mining and business intelligence.