2010:05  

Fuzzy Relations 
 

                    Majid Hussain 
 

Thesis for the degree Master of Science (two years)

in Mathematical Modelling and Simulation
30 credit points (30 ECTS credits)

May 2010
 

Blekinge Institute of Technology

School of Engineering
Department of Mathematics and Science
Supervisor: Elisabeth Rakus-Andersson

 
 
 
 

ii 
 
 
 

Abstract

One of the most fundamental notions in pure and applied sciences is the concept of a
relation. Science has been described as the discovery of relations between objects, states and
events. Fuzzy relations generalize the concept of relations in the same manner as fuzzy sets
generalize the fundamental idea of sets. This work presents an overview of comparison between
classical and fuzzy relations. Some important compositions of fuzzy relations have been
described and using these compositions a model for predicting score in cricket is developed.
Finally it deals with the restoration and the identification of the causes (diagnosis) through the
observed effects (symptoms) on the basis of fuzzy relations.

iii 
 
 
 

iv 
 
 
 

Acknowledgements
All praise for almighty Allah who guides us in darkness and help us in all difficulties and
problems. My sincere appreciation goes to my supervisor Prof.Elisabeth-Rakus Andersson for
her encouragement, continuous guidance, suggestion and keen interest in my thesis work. I
would like to thank programme manager DR. Raisa Khamitova for her unusually great help
throughout my study period. I wish to express my thanks to all my other teachers for their proper
guidance and friendly behaviour during my studies.

This all is the fruit of untiring efforts, lot of prayers, encouragement, sacrifices, guidance,
moral and financial support of my great, respectable and loving parents. I have no words to thank
my parents. I wish to thank my entire extended family for providing a loving environment for
me.

v 
 
 
 

vi 
 
 
 

Table of Contents
List of figures………………………………………………………………………………………………..

1 Comparison of classical relation and fuzzy relation

1.1 Introduction ...................................................................................................................................... 1
1.2 Crisp Relation .................................................................................................................................. 2
1.3 Fuzzy Relation ................................................................................................................................. 4
1.4 The Maximum-Minimum Composition of Relations ……………………………………………..5

1.5 Fuzzy Max-Min composition Operation……………………………………………………….......7

1.6 conclusion……………………………………………………………………………………….....9

2 Properties of fuzzy relations

2.1 Projection of Fuzzy Relation ......................................................................................................... 10
2.2 Cylindrical Extension of Fuzzy Relation ....................................................................................... 11
2.3 Reflexive Relation ....................................................................................................................... 11
2.4 Antireflexive Relation .................................................................................................................... 12
2.5 Symmetric Relation………………………………………………………………………………12
2.6 Antisymmetric Relation ................................................................................................................ 12
2.7 Transitive Relation ............................................................................................................. ………13
2.8 Similarity Relation………………………………………………………………………………..14
2.9 Antisimilarity Relation……………………………………………………………………………16
2.10 Weak Similarity Relation…………………………………………………………………………17
2.11 Order Relation…………………………………………………………………………………….17
2.12 Pre Order Relation………………………………………………………………………………..18
2.13 Half Order Relation................................................................... .. ………………….. …………..18

3 Applications of Fuzzy Relations

3.1 Fuzzy Graph ................................................................................................................................... 19
3.2 Complement of a Fuzzy Graph ...................................................................................................... 20
3.3 Model for Predicting Score in Cricket……………………………………………………………21
3.4 The Modus Ponens Law in Medical Diagnosis ............................................................................. 24
3. 5 conclusion………………………………………………………………………………………..27
References ………………………………………………………………………………………..28

vii 
 
 
 

List of Figures

1.1 Relation “equal to” and its characteristic function........................................................................... 4

3.1 Fuzzy graph.................................................................................................................................... 22
3.2 Complement of a fuzzy graph ........................................................................................................ 23

viii 
 
 
 
1.1 Introduction
In 1965, L. A. Zadeh introduced the concept of fuzzy set theory. Fuzzy set theory is an
extension of classical set theory. A logic that is not very precise is called a fuzzy logic. The
imprecise way of looking at things and manipulating them is much more powerful than precise
way of looking at them and then manipulating them. Fuzzy logic is one of the tools for making
computer system capable of solving problems involving imprecision. Fuzzy logic is an attempt to
capture imprecision by generalizing the concept of set to fuzzy set.

In every day content most of the problems involve imprecise concept. To handle the
imprecise concept, the conventional method of set theory and numbers are insufficient and need
to be extended to some other concepts. Fuzzy concept is one of the concepts for this purpose.

A relation is a mathematical description of a situation where certain elements of sets are

related to one another in some way. Fuzzy relations are significant concepts in fuzzy theory and
have been widely used in many fields such as fuzzy clustering, fuzzy control and uncertainty
reasoning. They also play an important role in fuzzy diagnosis and fuzzy modeling. When fuzzy
relations are used in practice, how to estimate and compare them is a significant problem.
Uncertainty measurements of fuzzy relations have been done by some researchers. Similarity
measurement of uncertainty was introduced by Yager who also discussed its application.

1 
 
 
 
1.2 Crisp Relation
To describe the fuzzy relation, first we describe relation by an example of daily life using
discrete fuzzy sets. Relationship is described between the colours of a fruit X
and the grade of maturity Y . Crisp set X with three linguistic terms is given as
X = {green, yellow, red}
Similarly the grade of maturity for the other set Y will be
Y = {verdant, half-mature, mature}
Crisp formulation of a relation X  Y between two crisp sets is presented in tabular form 

                                 
  Verdant Half- Mature
mature
Green 1 0 0
Yellow 0 1 0
Red 0 0 1

In the above table “0” and “1” describe the grade of membership to this relation. This relation is
a new kind of crisp set that is built from the two crisp base set X and Y . This new set is now
called R and can be expressed by the rules
1. If the colour of the fruit is green then the fruit is verdant.
2. If the colour of the fruit is yellow then the fruit is half-mature.
3. If the colour of the fruit is red then the fruit is mature.
This crisp relation shows the existence or absence of connection, relations or interconnection
between two sets. Now we show the membership grades represented in the fuzzy relation.
Verdant Half-mature Mature
Green 1 0.6 0
Yellow 0.4 1 0.3
Red 0 0.5 1

The table above represents the fuzzy relation.

2 
 
 
 
Crisp relation is defined on the Cartesian product of two universal sets determined as

X  Y  {( x, y) | x  X , y  Y }
    The crisp relation R is defined by its membership function

1,  x, y   R
 R  x, y   
0,  x, y   R

Here “1” implies complete truth degree for the pair to be in relation and “0” implies no relation.
When the sets are finite the relation is represented by a matrix R called a relation matrix.

1.2.1 Example
Let  X  1, 4,5 and  Y  3, 6, 7

              Classical matrix for the crisp relation when   R  x  y  is 
3 6 7
11 1 1 
R  4  0 1 1 
5  0 1 1 

1.2.2 Example
Let A  2, 4,6,8 and B  2, 4,6,8

 Classical matrix for the crisp relation   R  x  y  

2 4 6 8
2 1 0 0 0
4 0 1 0 0 
R
6 0 0 1 0
 
8 0 0 0 1

3 
 
 
 
 

Figure 1.1 Relation “equal to” and its characteristic function

1.3 Fuzzy relation

Let  X , Y  R  be universal sets then;  

R   x, y  ,  R  x, y   |  x, y   X  Y    
                         is called a fuzzy relation in  X  Y  R  

                or X  and  Y are two universal sets, the fuzzy relation R  x, y   is given as

     x, y  
R  x, y    R |  x, y   X  Y 
  x, y  
Fuzzy relations are often presented in the form of two dimensional tables. A m  n matrix
represents a contented way of entering the fuzzy relation R . 
y1  yn
x1   R  x1 , y1    R  x1 , yn  
 
R      
xm   R  xm , y1    R  xm , yn  


1.3.1 Example
Let  X  1, 2,3 and  Y  1, 2  

 If the membership function associated with each order pair  x, y   is given by  

                                                            R  x, y   e  x  y        
2

4 
 
 
 
 then derive fuzzy relation.
 
Solution 

The fuzzy relation can be defined in two ways using the standard nomenclature we have. 
 e112 e1 22 e  212 e 2 22 e 312 e 322 
R , , , , , 

 1,1 1, 2   2,1  2, 2   3,1  3, 2  
                  
 1.0 0.37 0.37 1.0 0.02 0.37 
  R , , , , ,  
 1,1 1, 2   2,1  2, 2   3,1  3, 2  
In the second method using the relational matrix, we have  
 1 0.37 

R   0.37 1   
 0.02 0.37 

Thus the membership function describes the closeness between set  X  and Y  . From 

the relational matrix it is obvious that higher values imply stronger relation.  

1.4 The maximum-minimum composition of relations

Let X ,  Y and  Z  be universal sets and let  R be a relation that relates elements from 

X  to Y , i.e. 

                  R    x, y  ,  R  x, y   x  X , y  Y , R  X  Y        

             and  

               Q  y, z  ,  Q  y, z   y Y, z  Z,Q  Y  Z  

Then  S will be a relation that relates elements in X that  R contains to the elements 

in  Z  that  Q contains, i.e. 
                                               S  R  Q                              
Here “  ” means the composition of membership degrees of  R and  Q  in the max‐min 
sense. 

S   x, z  ,  s  x, z   x  X , z  Z, S  X  Z  

  max‐min composition is then defined as 

5 
 
 
 

 
   S  x, z   max min  R  x, y  , Q  y, z 
yY
  
 and max product composition is then defined        


 S  x, z   max min   R  x, y  .Q  y, z  
yY
 

1.4.1 Example

Let  X   x1 , x2  and Y   y1 , y2  and Z   z1 , z2   

0 1  0 1 
                            R        and    Q    
1 0  1 0 

Then find the max‐min composition and max product composition 

                                                                 S  R  Q  

1 0 
S       is the max‐min composition. 
                                                                  0 1 

1 0 
                                and S  
                       0 1 
                           is the max product composition 
For crisp relations max‐min composition and max product will yield the same result, when 
X  has three elements, Y  has four elements and Z has two elements like    

X   x1 , x2 , x3      Y   y1 , y2 , y3, y4      Z   z1 , z2  then for relations

y1 y2 y3 y4
x1  1 0 1 0 
R  x2  0 0 0 1   
x3  0 0 0 0 

6 
 
 
 
x1 x2
y1  0 1 
                                                                            y2  0 0   
Q 
y3  0 1 
 
y4  0 0 

the max‐min composition is  
z1 z2
x1 0 1
                                                                             
S  x2 0 0 
x3 0 0 

In this example max‐min composition and max product have the same result. 

1.5 Fuzzy max-min composition operation

Let us consider two fuzzy relations R1 and R2 defined on a Cartesian space X  Y

and Y  Z respectively. The max-min composition of R1 and R2 is a fuzzy set defined on a

cartesian spaces  X  Z as

 
R1  R2   x, z  ,max min
 
 R1 R2  

 x, y  ,   y, z   | x  X , y Y , z  Z
where  R1  R2 is the max‐min composition of fuzzy relations  R1  and  R2  and max product composition is 

defined as  

 R R  max    x, y  .  y, z   | x  X , y  Y , z  Z
1 2  R1 R2 

1.5.1 Example
Let R1  x, y  and R2  x, y  be defined as the following relational matrices

0.6 0.5 
0.7 0.3 0.4
R1   1 0.1 and R2   
 0 0.7  0.9 0.1 0.6

We shall first calculate the max-min composition R1  R2

7 
 
 
 
0.6 0.5 
0.7 0.3 0.4 
R1  R2   1 0.1  

 0.9 0.1 0.6   
 0 0.7 

Now we calculate  

 R R  x1, z1   max  min  0.6, 0.7  , min  0.5, 0.9   max  0.6, 0.5  0.6
1 2

Similarly we can calculate the other entries. The relational matrix for max-min composition
in fuzzy relation is thus
0.6 0.3 0.5
R1oR2  0.7 0.3 0.4 
0.7 0.1 0.6 

1.5.2 Example
Let R1  x, y  and R2  x, y  be defined by the following relational matrix

y1 y2 y3 y4 y5
x1  0.1 0.2 0 1 0.7 
R1  x2  0.3 0.5 0 0.2 1 
x3 0.8 0 1 0.4 0.3 

z1 z2 z3 z4
y1 0.9 0 0.3 0.4 
y2 0.2 1 0.8 0 

R2  y3  0.8 0 0.7 1 
 
y4 0.4 0.2 0.3 0 
y5  0 1 0 0.8 

we shall first compute the max-min composition  R1  R2  x, z 

 R R  x1, z1   max  min  0.1, 0.9 , min  0.2, 0.2 , min  0, 0.8 , min 1, 0.4 , min  0.7, 0                         
1 2

 max  0.1, 0.2, 0, 0.4, 0   0.4

Similarly we can determine the grades of membership for all pairs

 xi , zi  , i  1, 2,3, j  1,4

8 
 
 
 
z1 z2 z3 z4
x1  0.4 0.7 0.3 0.7 
R1oR2  x2  0.3 1 0.5 0.8 
x3  0.8 0.3 0.7 1 

for the max product composition, we calculate

R1 
x1 , y1  .
R2 
 y1 , z1   0.1.0.9  0.09  

R1 
x1 , y2  .
R2 
 y2 , z1   0.2.0.2  0.04  

R1 
x1 , y3  .
R2 
                                                         y3 , z1   0.0.8  0  

                                                       
R1
 x1 , y4  . R  y4 , z1   1.0.4  0.4
2

R1 
 x1 , y5  .  y5 , z1   0.7.0  0
R2
hence

R 1o R 2 
 x1, z1   max 0.09, 0.04, 0, 0.4, 0  0.4

In the similar way after performing the remaining computation, we obtain

z1 z2 z3 z4
x1  0.4 0.7 0.3 0.56 
R1oR2  x2 0.27 1 0.4 0.8 
x3  0.8 0.3 0.7 1 

1.6 Conclusion
It is clear from the example that max-min composition and max product composition of
crisp relations will yield the same result, but in fuzzy max-min composition and max product
composition have different result.

9 
 
 
 
2.1 Projection of Fuzzy Relation
Let R    x, y  ,  R  x, y   |  x, y   X  Y  be a fuzzy relation. The projection of R  x, y 
on X denoted by R1 is given by

 
R1   x, max  R  x, y   |  x, y   X  Y 
 y  
and the projection of R  x, y  on Y denoted by R2 is given by

R2   y, max   x, y  |  x, y   X  Y 
x
R

2.1.1 Example

Let R be a fuzzy relation defined by the following relational matrix

y1 y2 y3 y4 y5 y6
x1  0.1 0.2 0.4 0.8 0.8  1
R  x2  0.2 0.4 0.8 1 0.8 0.6 
x3  0.4 0.8 1 0.8 0.4 0.2 

The projection of R  x, y  on X is calculated as, e.g.


R1
 x1   max 0.1, 0.2, 0.4, 0.8,1, 0.8  1
In the similar way can calculate the grades of membership for all pairs, so the X projection is
R 1  x1 ,1 ,  x2 ,1 ,  x3 ,1

The projection of R  x, y  on Y is calculated as, e.g.


R1
 y1   max 0.1, 0.2, 0.4  0.4
In the similar way we can determine the membership grade for all other pairs, so the Y
projection

R2  y 1, 
0.4  ,  y2 , 0.8  ,  y3 ,1 ,  y4 ,1 ,  y5 ,1 ,  y6 , 0.8 

10 
 
 
 
2.2 Cylindrical extension of fuzzy relation
The cylindrical extension on X  Y of a fuzzy set A of X is a fuzzy relation cylA whose
membership function is equal to

cylA  x, y   A  x  , x X,  y Y

Cylindrical extension from X-projection means filling all the columns of the related matrix by the
X -projection. Similarly cylindrical extension from Y projection means filling all the rows of
the relational matrix by the Y -projection.
2.2.1 Example
The cylindrical extension of R2 form the previous example is

y1 y2 y3 y4 y5 y6
x1 0.4 0.8 1 1 1 0.8
R2  x2 0.4 0.8 1 1 1 0.8
x3 0.4 0.8 1 1 1 0.8

2.3 Reflexive Relation

Let R be a fuzzy relation in X  X then R is called reflexive, if
 R  x, x   1 x  X

2.3.1 Example
Let X  1, 2,3, 4

1 2 3 4
1  1 0.9 0.6 0.2 
2 0.9 1 0.7 0.3
R 
3 0.6 0.7 1 0.9 
 
4 0.2 0.3 0.9 1 

is reflexive relation

11 
 
 
 
2.4 Antireflexive relations
Fuzzy relation R  X  X is antireflexive if
 R  x, x   0, x  X

2.4.1 Example
x1 x2 x3
x1  0 0 0.6 
is antireflexive relation
R1  x2 0.3 0 0 
x3  0 0.3 0 

2.5 Symmetric Relation

A fuzzy relation R is called symmetric if,

 R  x, y    R  y, x  x, y  X

2.5.1 Example
Let X   x1 , x2 , x3 

x1 x2 x3
x1  0.8 0.1 0.7 
is a symmetric relation.
R  x2  0.1 1 0.6 
x3 0.7 0.6 0.5 

2.6 Antisymmetric Relation

Fuzzy relation R  X  X is antisymmetric iff

if  R  x, y   0 then  R  y, x   0 x, y  X , x  y

12 
 
 
 
2.6.1 Example

x1 x2 x3
x1  0 0 0.7 
 is antisymmetric relation.
R  x2 0.2 0 0 
x3  0 0.2 0 

2.7 Transitive Relation

Fuzzy relation R  X  X is transitive in the sense of max-min iff

 R  x, z   max  min   R  x , y  ,  R  y , z    x , z  X
y X

since R 2  R  R if
 2  x, z   max   R  x, y  , R  y, z  
R yX

then R is transitive if R  R  R  R  R  R 

and R 2  R means that  R 2  x, y    R  y , x 

2.7.1 Example
Let X   x1, x2, , x3 

 0.7 0.9 0.4 

is R   0.1 0.3 0.5 a transitive relation?
 0.2 0.1 0 

Solution
 0.7 0.9 0.4  0.7 0.9 0.4 
R  R   0.1 0.3 0.5   0.1 0.3 0.5
 0.2 0.1 0   0.2 0.1 0 

0.7 0.7 0.5

R   0.2 0.3 0.3
2

 0.2 0.2 0.2 

13 
 
 
 
Since 
R
2  xi , x j  is not always less than or equal to  R  xi , x j  , hence R is not transitive.

2.7.2 Example
Let X   x1, x2, 

 0.4 0.2
is R    a transitive relation?
0.7 0.3

Solution
0.4 0.2 0.4 0.2
RR     0.7 0.3
 0.7 0.3   

using max-min composition

 max  min  0.4, 0.4  , min  0.2, 0.7   max  min  0.4, 0.2  , min  0.2, 0.3  
R2   
 max  min  0.7, 0.4  , min  0.3, 0.7   max  min  0.7, 0.2  , min  0.3, 0.3  

 max(0.4,0.2) max(0.2, 0.2) 

R2   
 max(0.4,0.3) max(0.2,0.3) 

0.4 0.2
R2   
0.4 0.3

 R  xi , x j  is less than or equal to  R  xi , x j  , so R is transitive.

2

2.8 Similarity Relations

R  X  X which is reflexive, symmetric and transitive is called the similarity relation.
x1 x2 x3 x4 x5 x6
x1  1 0.2 1 0.6 0.2 0.6 
x2 0.2 1 0.2 0.2 0.8 0.2 

x3  1 0.2 1 0.6 0.2 0.6  is a similarity relation.
R  
x4 0.6 0.2 0.6 1 0.2 0.8 
x5 0.2 0.8 0.2 0.2 1 0.2 
 
x6 0.6 0.2 0.6 0.8 0.2 1 

14 
 
 
 
2.8.1 Theorem
Each equivalence class R  X  is given as

R X     ,    0,1

R  X 

where R  X  is the -cut of R   .

2.8.2 Definition
A  X , A is a fuzzy set the  -cut of A is a non fuzzy set denoted by A and defined by

A   x :  A  x     ,    0,1

2.8.3 Example
For R  x1  we have

R0.2  x1    x1 , x2 , x3 , x4 , x5 , x6 

0.2  0.2  0.2  0.2  0.2  0.2  0.2

R0.2  x1  x1 x2 x3 x4 x5 x6

R0.6  x1    x1 , x3 , x4 , x6 

0.6  0.6  0.6  0.6  0.6

R0.6  x1  x1 x3 x4 x6

R1  x1    x1 , x3 

1  1  1
R1  x1  x1 x3

Equivalence class for  R  x1   

R  x1   0.2  0.2  0.2  0.2  0.2  0.2  0.6  0.6  0.6  0.6  1  1
x1 x2 x3 x4 x5 x6 x1 x3 x4 x6 x1 x3
max  0.2, 0.6,1 max  0.2, 0.6,1 max  0.2, 0.6  max  0.2, 0.6 
R  x1    0.2    0.2 
x1 x2 x3 x4 x5 x6
R  x1   1  0.2  1  0.6  0.2  0.6
x1 x2 x3 x4 x5 x6

15 
 
 
 
2.8.4 Example

Equivalence class for the similarity relation R is

R  x1   1  0.2  1  0.6  0.2  0.6

x1 x2 x3 x4 x5 x6

R  x2   0.2  1  0.2  0.2  0.8  0.2

x1 x2 x3 x4 x5 x6

R  x3   1  0.2  1  0.6  0.2  0.6

x1 x2 x3 x4 x5 x6

R  x4   0.6  0.2  0.6  1  0.2  0.8

x1 x2 x3 x4 x5 x6

R  x5   0.2  0.8  0.2  0.2  1  0.2

x1 x2 x3 x4 x5 x6

R  x6   0.6  0.2  0.6  0.8  0.2  1

x1 x2 x3 x4 x5 x6

2.9 Antisimilarity Relation

If R is a similarity relation then the complement of R is antisimilarity relation.    
R  X  X  is a antisimilarity relation if 

 R /  x, y   1   R  x, y 
The antisimilarity relation is antireflexive, symmetric and transitive in the sense of max-
min, i.e. 

 R /  x, z   min  max

yX 

 R /  x, y  ,  R /  y, z   x, z  R 
2.9.1 Example
 1 0.1 0.7 
Prove that R   0.1 1 0.7  is antisimilarity relation?
0.7 0.7 1 

16 
 
 
 
Solution
According to definition of antisimilarity relation

 R  x , y   1   R  x, y 
/

 1 0.1 0.7 
 R /  x, y   1   0.1 1 0.7
0.7 0.7 1 

 0 0.9 0.3
 R /  x, y   0.9 0 0.3  
 0.3 0.3 0 
                 

This is anti-reflexive, symmetric and transitive, so R is antisimilarity relation.

 

2.10 Weak Similarity  

R  X  X  which is reflexive and symmetric is called the relation of weak similarity 
(not transitive). 
 1 0.1 0.8 0.2 0.3
 0.1 1 0 0.3 1 

R   0.8 0 1 0.7 0  is weak similarity relation
 
0.2 0.3 0.7 1 0.6 
 0.3 1 0 0.6 1 

2.11 Order Relation

An order relation R  X  X is transitive relation in the sense of max-min; i.e

 R  x, z   max  min   R  x , y  ,  R  y , z    , x, z  X
y X

17 
 
 
 
2.12 Pre Order Relations
A pre order relation R  X  X is reflexive and transitive in the max-min sense e.g.
x1 x2 x3 x4 x5
x1  1 0.7 0.8 0.5 0.5
x2  0 1 0.3 0 0.2 

R  x3  0 0.7 1 0 0.2 
 
x4 0.6 1 0.9 1 0.6 
x5  0 0 0 0 1 

2.13 Half Order Relation

A fuzzy half order is a relation R  X  X which is reflexive
 R  x, x   1 x  X

and weakly antisymmetric, i.e.

if  R  x, y   0 and  R  y, x   0 then x  y

x1 x2 x3 x4 x5 x6
x1 1 0.8 0.2 0.6 0.6 0.4
x2 0 1 0 0 0.6 0 

x3 0 0 1 0 0.5 0  is half order relation
R  
x4 0 0 0 1 0.6 0.4 
x5 0 0 0 0 1 0 
 
x6 0 0 0 0 0 1 

18 
 
 
 
3.1 Fuzzy Graph
In 1975, Rosenfeld considered fuzzy relations on fuzzy sets. He developed the theory of
fuzzy graphs. Bang and Yeh during the same time introduced various connectedness concepts in
fuzzy graph. Inexact information is used in expressing or describing human behaviors and mental
process. The information depends upon a person subjectively and it is difficult to process
objectively.

Fuzzy information can be analyzed by using a fuzzy graph. Fuzzy graph is an expression
of fuzzy relation and thus the fuzzy graph is frequently expressed in fuzzy matrix.

Mathematically a graph is defined as G  V , E  where V denotes the set of vertices

and E denotes the set of edges. A graph is called a crisp graph if all the values of arcs are 1 or 0
and a graph is called fuzzy graph if its values is between 0 and 1.Fuzzy graph G   ,   is a

pair of functions  : S   0,1 where S is the set of vertices and  : S  S   0,1 , x, y  S .

Fuzzy graph H   , v  is called a fuzzy subgraph of G if

  x     x  , x  S and v  x, y     x, y  x, y  S

3.1.1Example
Fuzzy relation is defined by the following fuzzy matrix the corresponding fuzzy graph is
shown in the fig (3.1)

b1 b2 b3
a1  0.5 1.0 0.0 
a2 0.0 0.0 0.5
a3 1.0 1.0 0.0 

19 
 
 
 
a1
0.5 b1

1.0

a2 1.0 b2
0.5
1.0

a3 b3

Fig 3.1

Fuzzy graph

3.2 Complement of a Fuzzy Graph

The complement of a fuzzy graph G :  ,   is a fuzzy graph G : ( ,  ) where    and

 (u , v )   (u )   ( v )   ( u , v )  u , v  V

Complement of a fuzzy graph are shown in fig below

(0.5)u1 u2 (0.6) (0.5)u1 u2 (0.6)

(0.4)u4 u3 (0.3) (0.4)u4 u3 (0.3)

G G

Fig 3.2( a )

Complement of a fuzzy graph

20 
 
 
 

(0.6)v1 v2 (0.4) (0.6)v1 v2 (0.4)

(0.4)v4 v3 (0.6) (0.4)v4 v3 (0.6)

G G

Fig 3.2  b 

Complement of a fuzzy graph

3.3 Model for Predicting Score in Cricket

In this model we can predict score using max-min composition,max product composition
and max-av composition.
Speed of bowling = fast bowling, medium bowling, spin bowling and
Y= condition on pitches= {good wicket, fair wicket, sporting wicket, green wicket, crumbling
wicket, rough wicket}
Let R denotes the relationship between speed of bowling and condition on pitch and Q
denotes the relationship between conditions on pitches and runs on the board.

gd .w f .w s.w gr.w c.w r.w

fast  0.6 0.5 0.4 0.1 0.9 0.5 
R  medium  0.8 0.6 0.9 0.2 0.1 0.6 
spin  0.7 0.8 0.6 0.7 0.1 0.2 

21 
 
 
 
low.r ave.r hig.r
gd .w  0.4 0.8 0.7 
f .w  0.3 0.8 0.8 
and s.w  0.2 0.7 0.8 
Q  
gr.w  0.8 0.6 0.4 
c.w  0.7 0.5 0.4 
 
r.w  0.9 0.4 0.2 

R  Q  Relationship between speed of the bowling and runs on the board

We calculate R  Q by using max-min composition rule

max min(0.6, 0.4), min(0.5, 0.3), min(0.4, 0.2), min(0.1, 0.8), min(0.9, 0.7), min(0.5, 0.9)

 max 0.4, 0.3, 0.2, 0.1, 0.7, 0.5

 0.7
Similarly we can calculate the other entries
The relational matrix for max-min composition in fuzzy relational is thus

low.r ave.r hig .r

fast  0.7 0.6 0.6 
R  Q  medium  0.6 0.8 0.8  (3.1)
spin  0.7 0.8 0.8 

Max Product composition

Now by using max product composition we find the relationship between speed of the
bowling and runs on the board
R  Q  Relationship between speed of the bowling and runs on the board
We calculate R  Q by using max product composition rule

max  0.6.0.4, 0.5.0.3, 0.4.0.2, 0.1.0.8, 0.9.0.7, 0.5.0.9 

 max  0.24, 0.15, 0.08, 0.08, 0.63, 0.45 

 0.63

22 
 
 
 
Similarly
max  0.48, 0.4, 0.28, 0.06, 0.45, 0.2 

 0.48
and
max  0.42, 0.4, 0.32, 0.04, 0.36, 0.1

 0.42

Similarly we calculate the other entries and the relational matrix for max product composition is
low.r ave.r hig .r
fast  0.63 0.48 0.4 
R  Q  medium  0.54 0.64 0.64  (3.2)
spin  0.56 0.64 0.64 

Max-av Composition
Now by using max product composition we find the relationship between speed of the
bowling and runs on the board

R  Q  Relationship between speed of the bowling and runs on the board

We calculate R  Q by using max-av composition rule
1
.max  0.6  0.4,0.5  0.3,0.4  0.2,0.1  0.8, 0.9  0.7,0.5  0.9 
2
1
 .max 1, 0.8, 0.6, 0.9, 0.16, 0.14 
2
1
  0.16 
2
 0.8
for the second entry
1
.max  0.14, 0.13, 0.11, 0.7,0.14, 0.9 
2
1
  0.14 
2

 0.7

23 
 
 
 
for third entry
1
.max  0.13, 0.13, 0.12, 0.5,0.13, 0.7 
2
1
  0.13
2
 0.65
Similarly we calculate the other entries and the relational matrix for max-av composition is
low.r ave.r hig .r
fast  0.8 0.7 0.65 
R  Q  medium  0.85 0.8 0.85 

(3.3)
av
spin  0.75 0.8 0.8 

By analyzing the results of  3.1 ,  3.2  and  3.3 we conclude that  3.2  is more reliable.

3.4 The Modus Ponens Law in Medical Diagnosis

The creators of fuzzy set theory, who develop mathematical models applied to different
technical domain, have also made representative contributions in medical investigation. To
decide an appropriate diagnosis in one patient we introduce three non fuzzy sets
The set of symptoms S  S1 , S2 ,, Sm 

The set of diagnosis D   D1 , D2 , , DP 

The set of patients P   P1 

The symptoms occurring in set S are associated with the diagnosis from set D .The symptoms
S1 , S2 ,, Sn that are stated in set S are included in the pairs  P1 , S1  ,  P1 , S2  , ,  P1 , Sn  .Fuzzy

relation PS as a one row matrix

S1 S2 ... Sn
PS    PS  P1, S1   PS  P1, S2  ...  PS  P1, Sn  

where  PS  P1 , S j  , j  1, 2, , n is a value of the membership degree providing us with

evaluation of the intensity S j in P1 .

The next relation consists of the pairs  S1 , D1  ,  S1 , D2  ,,  Sn , DP  .

24 
 
 
 
S1   SD  S1 , D1    SD  S1 , DP  
 
SD       
S n   SD  S n , D1    SD  S n , DP  

The fuzzy relation in which each value of the membership degree tied to the pair

S j , Dk  , j  1, 2,  , n k  1, 2,  , P and patient-diagnosis relation is a matrix

D1 D2 DP
PD    PD  P1 , D1   PD  P1 , D2  ...  PD  P1 , DP  

Finally the relation which allowing us to estimate association between the patient and the
considered diagnosi

PD  PS  SD

3.4.1Example

In this example we can find an appropriate diagnosis using relation between patient to
symptoms and symptoms to diagnosis. To find this we can choose

The set of patients P   P1 

The set of symptoms S  S1 , S2 ,, S4 

= {smoking, hypertension, increased level of LDL-cholesterol, pain in chest}

The set of diagnosis D   D1 , D2 , D3 

The set of diagnosis D  {high risk of cardiovascular diseases, coronary heart diseases,           
myocardial infarct}
The purpose is to find one of diagnosis in P1 who shows the presence of symptoms in a certain

degree. PS formed by the questions (asked by a physician or by a questionnaire).

PS   0.913 0.81 0.63 0.2

Now we calculate symptoms-diagnosis relation by introducing linguistic variables

Presence = {never, almost never, very seldom, seldom, rather seldom, moderately, rather often,
often, very often, almost always, always}

25 
 
 
 
Numerical description of fuzzy variables in “presence”

Fuzzy variables x " common "  x 

“never” 7.5 0
“almost never” 15 0.016
“very seldom” 22.5 0.062
“seldom” 30 0.14
“rather seldom” 37.5 0.25
“moderately” 50 0.5
“rather often” 62.5 0.75
“often” 70 0.86
“very often” 77.5 0.938
“almost always” 85 0.984
“always” 92.5 1

So SD can be constructed as
 0.75 0.984 0.14 
0.86 0.5 0.938 
SD   
0.14 0.5 0.984 
 
 0.25 0.984 0.86 
The relation PD represents
PD  PS  SD
 0.75 0.984 0.14 
0.86 0.5 0.938
PD   0.913 0.81 0.63 0.2   
0.14 0.5 0.984 
 
 0.25 0.984 0.86 

by using max-min composition rule

max min  0.913, 0.75 , min  0.81, 0.86  , min  0.63, 0.14  , min  0.2, 0.25

26 
 
 
 
 max 0.75, 0.81, 0.14, 0.2

 0.81
Similarly we can calculate the other entries
D1 D2 D3
PD   0.81 0.913 0.81

According to the rule, the higher degree the more probable diagnosis, we choose D2

3.5 Conclusion
Fuzzy relations generalize the concept of fuzzy sets to multidimensional universes and
introduce the notion of association degree between the elements of some universe of discourse.
Fuzzy relations generalize the concept of relations in the same manner as fuzzy sets generalize
the fundamental idea of sets. Operations with fuzzy relations are important to process fuzzy
models constructed via fuzzy relations. Relations are associations and remain at the very basis of
most methodological approaches of science and engineering. Fuzzy relations are more general
constructs than functions; they allow dependencies between several variables to be captured
without necessarily committing to any particular directional association of the variables being
involved.

27 
 
 
 
References

1 H.J.Zimmermann “Fuzzy set theory and its application”, 4th edition, Kluwer Academic
Publishers, Boston, 2001.

 2 A.Rosenfeld and Fuzzy graphs. “In Fuzzy sets and their application to cognitive and
decision processes” L.A.Zadeh, K.S.Fu, K. Tanaka and M.Shimura, Academic Press New
York, 1975.
3 Witold Pedrycz and F.Gomide “Fuzzy Systems Engineering towards Human-Centric
Computing”, John Wiley and sons, Inc, 2007

 4 Raymond Yeh and S.Y.Bang Fuzzy Relations, Fuzzy Graphs and Their Application to

Clustering analysis, “In Fuzzy sets and their application to cognitive and decision
processes” L.A.Zadeh, K.S.Fu, K. Tanaka and M.Shimura, Academic Press New York,
1975.
 5 Elie Sanchez “Eigen Fuzzy sets and Fuzzy Relations. Journal of Mathematical Analysis

and Applications” 811981 399  421

 6 K.P.Adlassing “Fuzzy Modeling and Reasoning in a Medical Diagnostic Expert system

EVD in Medizin und Biologie” 17 1980 

7 Elisabeth Rakus- Andersson, “Fuzzy and Rough techniques in Medical Diagnosis and

Medication”, Springer, Berlin Heidelberg, 2007.

28