See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/325713159

Exam Time Table Scheduling using Graph Coloring Approach

Article  in  International Journal of Computer Sciences and Engineering · May 2018

DOI: 10.26438/ijcse/v6i5.8493

CITATIONS READS
5 6,048

2 authors, including:

Rubul Kumar Bania

North Eastern Hill University, India, Tura Campus
20 PUBLICATIONS   129 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Handwritten Assamese Character Recognition using Texture and Diagonal Orientation features with Artificial Neural Network View project

All content following this page was uploaded by Rubul Kumar Bania on 12 June 2018.

The user has requested enhancement of the downloaded file.

 International Journal of Computer Sciences and Engineering Open Access
Research Paper Vol.-6, Issue-5, May 2018 E-ISSN: 2347-2693

Exam Time Table Scheduling using Graph Coloring Approach

Rubul Kumar Bania1*, Pinkey Duarah2

1*
Department of Computer Applications, North-Eastern Hill University, Tura Campus, Meghalaya, India
2
Department of Computer Applications, North-Eastern Hill University, Tura Campus, Meghalaya, India
*
Corresponding Author: [email protected]

Available online at: www.ijcseonline.org

18/May/2018, Published: 31/May/2018

Abstract— One of the most common academic scheduling problems which can be perceived in any educational system is the
exam time table generation. The presence of vast numbers of students and offered courses makes it difficult to schedule exams
in a limited epoch of time. An appropriate schedule can be designed by utilizing different resources like subjects, teachers,
students and classrooms in a way to evade conflicts by fulfilling special types of constraints. Graph coloring is one decent
approach which can deal with timetable scheduling problem and can satisfy changing requirements. In this work, we have
framed a systemic model by applying graph vertex coloring approach for generating exam timetabling with the help of a course
matrix generated from given data of an educational institute. From the problem domain, different types of constraints viz., hard
and soft are figured out and while solving emphasis is focused on the degree of constraint satisfaction. Workflow of the system
is described by using a case study and the output which it has generated is efficient and satisfactory.

Keywords— Time table, Graph coloring, Scheduling, Hard constraints, Soft constraints, Course matrix
I. INTRODUCTION the lab. We should also consider those students who have
back papers (repeater exams) and thus, need to reappear in
Exam time table is required in every educational institution.
some of the papers. The exam time table should avoid such
In every semester or year, the universities and colleges are
conflicts, like no two or three exams for the identical student
required to generate exam time table for conducting the
should be scheduled at the same period of time. The exam
internal and the final semester exams. The presence of a
schedule should not leave a big gap between exams for the
large number of students and large number of offered courses
students. Therefore, there is a much need of an effective and
sometimes makes it difficult to schedule the exam without
accurate timetable to the performance of any educational
having any conflict [1]. Moreover, the exam time table could
institute. Graph coloring [2][10][12] is a method of assigning
not be reused because the requirements and some restrictions
colors to certain elements of a graph subjected to certain
of the problem keep changing. If done manually, generating
an exam time table is very time consuming and requires a constraints. Thus, optimal solutions to such problem may be
considerable amount of workforce which sometimes may found by determining minimal colorings for the
lead to inefficiency. So it demands an automatic generation corresponding graph. Also, the time table scheduling
of time table with a limited number of user input parameters. problem has been classified as a NP-hard problem [1] [3].
The exam time table should be generated in such a way that That means it is unlikely to find a fast algorithm to solve this
one subject is scheduled only once. So this can be considered problem, and to find a solution to such problem we need to
as a good constraint or restriction which must be satisfied consider all the possible solutions, and then choose the best
while designing an automatic time table generation system. one.
There is no restriction regarding the faculty members. That
means any faculty member could be assigned to conduct the In this work, we have contributed to formulate the exam
examination of any subject. However, if a faculty member timetabling problem followed by a systematic model framed
conducts the examination of those subjects which he/she on two methods with some user input parameters. It will
teaches then it would be beneficial because sometimes there automatically generate an exam time table for universities
could be some printing mistakes or some other errors on the based on some specified hard and soft constraints using the
question paper, which could be identified and immediately graph coloring approach. We have tested the methods with
corrected by the teacher. The theory exams could be two case studies of varying size of data and constraints. This
conducted in any room accordingly to the availability of the system has achieved fairness and tries to minimize the exam
room. However, practical exams could be conducted only in time period.

© 2018, IJCSE All Rights Reserved 84

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

A. Basic concept of Graph and Graph coloring Here, the resource could be a faculty member, class room,
Definition 1: A graph G, is a non-empty finite set V of lab etc.
vertices (nodes) and a non-empty finite set E of edges (arcs)
[4] [5] [11]. So, a graph G can be represented as an ordered
of (V(G), E(G)) consisting of a non-empty set V of vertices
or nodes, where E ⊆VxV. An undirected graph could be
pictorially represented as Figure-1, where, V = {A, B, C},
and E = {(A, C), (B, C), (A, B)}. Here e1= (A, C), e2= (B,
C) and e3= (A, B).

Figure 2. (a) Vertex coloring (b) Edge coloring

The time table problem could be divided into three

categories:

1. Exam time scheduling- It involves scheduling the

exams for different subjects.
2. Class time table for Universities- It involves
Figure 1. An example of a Graph. creating monthly time table of the university.
3. Class time table for Schools- It involves creating
weekly time table of school.
Definition 2: The least number of colors necessary for vertex
coloring of a graph G is called the chromatic number of G. It So, after discussing the fundamental and introductory
is normally denoted as £(G) [4][5]. It holds two properties: concepts of graph related to coloring and scheduling
problem, in the above subsection A and B, we have presented
1. £(G) =1, iff G is a null graph. the literature survey in Section II, which describes the state-
2. If G is not empty, then £(G) >=2. of-the-art of several research studies that have been done on
scheduling problems using graph coloring method. In Section
Also, if there is a vertex coloring that uses at most n colors, III formulation of the problem is derived. Methodology with
then G is said to be n-coverable, it means £(G)<=n. a workflow diagram is discussed in Section IV. In Section V
a case study with a worked example is shown. Next Sections
Definition 3: Vertex coloring [5] [7] is the most commonly VI and VII summarize the conclusion and future work of the
used graph coloring approach. It is a systematic manner of present study.
coloring the vertices of a graph in such a way that no two
adjacent vertices have the same color. Two nodes are said to II. RELATED WORK
be adjacent if they are connected by an edge. So, vertex
coloring of a graph G can be defined as a function Ƒ: V → C Solving timetabling problems through the application of
where V is a set of vertices of the graph G and C is a set of modern technology has a long and varied history. Welsh et
colors. Proper Ƙ-coloring of G is a coloring function Ƒ which
al. [1] has illustrated the relationship between timetabling
uses exactly Ƙ colors and satisfies the property that Ƒ (x) ≠ Ƒ
(y), where (x,y) is adjacent to each other. Edge coloring on and graph coloring and developed a new general algorithm to
the other hand, is a systematic way of assigning colors to the give an approximate solution to the minimum coloring
edges of a graph so that no two adjacent edges have the same problem more efficiently. However, this algorithm needs the
color. Two edges are said to be adjacent if both of them share vertices to be sorted in descending/ascending order based on
a vertex in common. In Figure 2(a)-(b), vertex and edge their valence (it is the degree of a vertex) in order to have an
coloring are shown respectively. optimal solution. Mohammad et al. [2] have proposed an
B. Time Table Scheduling Problem algorithm to color the schedules of exam time table in two
The fundamental initiative for designing a optimized major steps. In the first steps, they have a created a weight
conflict-free schedule is to allocate various related resources matrix and an undirected graph, and then in the second step
amongst a number of time-slots fulfilling a range of types of assign color to different nodes of that undirected graph. By
essential and preferential constraints [3][6]. Exam time this way, they tried to achieve a fair, accurate, and optimal
scheduling is a part of the time table problem. The time table exam time period by considering different constraints which
problem could be viewed as a resource allocation problem. generally related with university systems. Burke et al. [3]

© 2018, IJCSE All Rights Reserved 85

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

also followed a similar approach like the exam scheduling then course_matrix[i][j]= 1, and if the edge (vi, vj) is not
algorithm using graph coloring. However, in this algorithm, present in E(G), then course_matrix[i][j] = 0. Also some
they have addressed only the conflicts without any others terms need to be mentioned:
constraints. Moreover, this algorithm does not eliminate
conflicts, and only aims at minimizing conflicts. Akbulut et
1. Let H be the set of all periods of a time slot where
al. [6] proposed a graph coloring algorithm and RFID
examination can be conducted. H= {h1, h2, h3,…, hm},
technology to schedule different exams in same halls where m corresponds to the maximum number of
simultaneously. This system plans to schedule different periods of timeslots in a day.
exams in same halls at the same time period. The reason is to 2. Let S be the set of all subjects in a given semester/year,
make use of the hall capacity with 100% performance and which will be under examination. S= {s1, s2, s3,…, sk},
decrease cheating attempts. It uses the identity cards of where k is the maximum number of subjects.
students that have a RFID tag for this purpose. Exam time 3. Let F be the set of faculties available in a given
tabling problem is not only solved by graph coloring semester examination. F= {f1, f2, f3,…, fk}, where k is
approaches but also some researcher has attempted the number of faculty.
evolutionary approaches [12]. Jha [9] has attempted to give a 4. Let R be the set of rooms available in a given semester
solution to the problem of timetabling, by using the genetic examination. R= {r1, r2, r3,…,rk}, where k is the
algorithm (GA). Another approach has been carried out by maximum number of rooms available.
5. Let L be the list of combinations of subjects offered to
Verma et al. [8], where author has proposed a method to
students. L= {L1, L2, L3,…, Li}, where Li
schedule timetable and generate it using a combined
={combination of subjects from set S}.
approach of bacterial foraging and genetic algorithm
techniques. But the computational cost of this approach is Also, we need a set of color list C= {c1, c2, c3,…, cn}, where
comparatively high. In a similar work, Hussain et al.[7] has n is the maximum number of colors. By using these
presented a method for exam timetabling problem for the terminologies our primary objective is to minimize the
number of colors for solving the problem.
centre of foundation studies and extension education,
Multimedia University, Malaysia. Here the authors have used A. Constraints
graph coloring with the combination of cluster heuristic and To designing and solving a scheduling problem, the essential
sequential heuristic approach for solving the problem. But things that are to be considered are the constraints [3]. They
this approach is not applied for automated or computerized are the diverse margins and limitations mixed up in creating a
schedule. Based on the fulfilment of constraints, a schedule
generation of timetable.
can be accepted or get rejected. Also in the literature,
depending on the degree of strictness, constraints are broadly
III. PROBLEM FORMULATION
classified into-two categories namely, hard and soft
Constraints [3]. The exam time table should be generated by
keeping in mind the number of students, the number of
A vertex in graph G represents a subject, and an edge faculty members available to conduct the exams, the
between two vertices is drawn only if there is a common availability of rooms, and the capacity of the available rooms.
student. Two vertices could have the same color if and only
Hard constraints are compulsory. That means the hard
if they are not adjacent to each other. Since they are not constraints must have to be satisfied for the time table to be
adjacent; the subjects represented by the two vertices could feasible. Soft constraints are those constraints, which are not
be scheduled at the same time slot. The weight of an edge, if compulsory to be satisfied. However, if satisfied, they make
present, represents the number of students who have the time table more acceptable.
registered for both the subjects. We need to find the
minimum number of colors required to color the graph (i.e. Some of the hard constraints (also known as essential time
the chromatic number). Thus, we need to find the shortest tabling conditions) are listed below:
period within which the exams of all the subjects could be
conducted without any conflict. So, by considering a graph G 1. Two or more subjects or papers of a student should not
= (V, E). The number of vertex V is equal to the number of be scheduled at the same time.
subjects and each node represents a subject. The Course 2. The examination for each subject should be held exactly
Matrix of G is a two-dimensional nxn array, say once.
course_matrix[n][n], where n is the total number of subjects
to be scheduled. Now, if the edge (vi, vj) is present in E(G),

© 2018, IJCSE All Rights Reserved 86

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

3. Subjects should be scheduled accordingly to the other if they are adjacent. Adjacent subjects could not be
availability of rooms. assigned the same color.
4. Subjects should be scheduled accordingly to the
availability of faculty members. For coloring the vertices, following steps are performed and
it is shown in Method-2.
Some of the soft constraints [3][2] (also known as
preferential time tabling conditions) are shown below:
Method-2:
1. There should not be any exam on Sunday or holidays. Step 1: Create an empty array Arr[n] of size n. Each index
2. There should be a gap between two exams of a represents the subjects and initialized the array to „0‟. Here, n
particular stream. = total no. of subjects.
3. Classrooms should be large enough to accommodate Step 2: Generate a color array C ={c1, c2, c3….ck}. Declare
the number of students for a particular subject. an array color_used[k].
Step 3: Assign c1 to Arr[1] . Then Store c1 to color_used[1] .
Step 4: For each Row [1to n] of course_matrix, excluding the
IV. METHODOLOGY
first Row [1].
In this section, various steps that need to be performed to Step 5: For each index i of Arr, excluding the first index.
schedule an exam timetable are discussed. The overall Perform the following steps.
workflow diagram for this work is shown in Figure 3. The Step 6: Assign new color from C to Arr[i] which is not
system would first take the inputs from the user. So, we need previously used
to enter the following information: Step 6.1 If Vi adjacent to Row [1 to i].
Step 6.2 Then Store color to color_used array.
1. The list of subjects or papers offered in each semester. Step 7: Else Assign color to non-adjacent uncolored vertices
2. The number of students registered in each paper. from color_used.
3. The number of faculties available to conduct the Step 8: Repeat step 4 to step 7 until all vertices are colored.
examinations.
4. The number of rooms available. Once the color values or colors are assigned, the list of
5. The capacity of each room should be entered. Each subjects will be sorted based on the color value assigned by
room is identified by a unique name; say an alphabet, using the Merge-sort algorithm. We sort the list to ease the
followed by the capacity of the room. Example, A50 task of generating the exam time schedule and eliminate the
where A is the unique name and 50 is the capacity of need for searching the list again and again. Finally, it would
the room. generate the exam time schedule and allocate the suitable
Then it generates the course matrix using Method-1.To room/rooms. The rooms would be allocated depending on the
generates the course matrix following steps is being number of students registered in the subjects who are going
performed: to be scheduled in the same time slot with available room
capacity, and the number of faculties available to conduct the
Method-1: exams.
Step 1: Create an empty nxn two dimensional array V. AN EXPERIMENTAL CASE STUDY
course_matrix[n][n].
Step 2: Pick a vertex Vi and find the adjacent vertex Vj to In India, generally in Undergraduate/Postgraduate
make an edge between them. colleges/universities offers a variety of subject combinations
to students. In streams like Bachelor of Arts (B.A)/Bachelor
Step 3: For finding adjacency of Vi , perform a searching in
of Science (B.Sc) or Master of Arts (M.A) /Master of
the list L of subjects where it presents.
Science (M.Sc) students can take some combinations of
Step 3.1 If Vi ϵ Li then all the subjects in Li are adjacent
subjects with elective and core papers. Also, some students
to Vi , so assign course_matrix[i][j]= 1
have some back papers of some previous semesters. So those
Step 3.2 else course_matrix[i][j] = 0.
papers also will be included while offering the subjects.
Step 4: Repeat the step 2 and step 3 until all the pairs of While designing the time table availability of the number of
subjects are not assigned with „0‟ or „1‟. faculties plus numbers of room availability also needed. In
Based on the course matrix, the system would detect the this section, we have considered a scenario of an individual
colliding subjects and assign the colors to the nodes, which department of Computer application, where different subjects
represent the subjects. Two courses are said to collide each are offered by a university in an odd semester of a particular
year. The number of papers or subjects offered in different

© 2018, IJCSE All Rights Reserved 87

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

semesters, number of faculties, numbers of students and 6. Number of faculties to conduct the examination is say,
number of available rooms are taken into consideration for seven (7).
generating a conflict and constraint free solution to final 7. Starting Date of the exams and Time slots. Starting date
semester exam timetable. : Dec-01, 2017. And Time slots 10 a.m. to1p.m. and 1.30
to 4.30 p.m.
User Inputs

A. Generating the Course Matrix

The next step is to generate the course matrix using Method -
Generate the Course Matrix 1, based on the data provided by the user. Here, we are
generating the course matrix based on the list of subjects
offered in each semester including the back papers of the
previous semesters. The course matrix generated for the
Color Assignment
above data is shown in Table-3 and the graph which it will
generate is shown in Figure 4.
B. Color Assignment
Sort the Subjects Based on the Assigned Color Value
We assign colors to the subjects based on the course matrix
using Method-2. In order to assign a color to subjects we
need to find out the colliding subjects. Two courses are said
Generate the Exam Time Table and Allocate the Rooms to collide each other if they are adjacent. Adjacent courses
could not be assigned the same color. In Table-1, if we
Figure 3. Workflow diagram of the methodology observe the first row we see that DL and C could not be
assigned the same color since they are adjacent. However,
Suppose we have entered the following information, with DL and CD could be assigned the same color since they are
core subjects, elective subjects and back-papers. not adjacent
1. First-Semester subjects: Digital Logic (DL), C Now, we take a one-dimensional array of size n, n = 14, it is
Programming (C), Organizational Behaviour (OB), shown in Table-1. Each index of the array represents the
Accounting (AC), Discrete Mathematics (MATH) subjects. Now initialized the array to 0, also a Color array is
2. Third-Semester subjects: Database Management System initialized: {1=Green, 2=Red, 3=Blue, 4=Black, 5=light-
(DBMS), Data Communications and Network orange, 6= light-purple, 7= Orange, 8= light-Blue 9=Pink,
Technologies (DCN), Operating System (OS), 10= Maroon} like that.
Algorithm (ALG), VB.NET (VB), DL, C, MATH. Table 1. Color assignment of Digital Logic (DL)
NOTE: DL, C, and MATH are the first semester
subjects. However, we are adding these subjects in the Subjects DL C OB AC MATH DBMS DCN OS ALG
third semester because we are assuming that some third Color 0 0 0 0 0 0 0 0 0
semester students have got back paper in these three
papers and thus need to reappear. Here, Algorithm VB CD SP DM JAVA
0 0 0 0 0
(ALG) is an elective paper.
3. Fifth-Semester subjects: Compiler Design (CD), System
Since, we start with Digital Logic (DL), we assign color 1 to
Programming (SP), Data Mining (DM), JAVA, C, ALG.
it. It is shown in Table-2.
NOTE: C, and ALG are first and third semester subjects
respectively. However, we are adding these subjects in the Table 2. Color assignment of Digital Logic (DL)
Fifth semester because we are assuming that some Fifth
semester students have back papers in these subjects and thus Subjects DL C OB AC MATH DBMS DCN OS ALG
Color 1 0 0 0 0 0 0 0 0
need to reappear. Here, Data Mining is an elective paper.
4. Enter the number of students enrolled in each subject. VB CD SP DM JAVA
5. Number of rooms available is 9 (e.g., A30, B30, C30, 0 0 0 0 0
D40, E40, F30, G40, H60, I40).

© 2018, IJCSE All Rights Reserved 88

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

Table 3. Course matrix

Subject DL C OB AC MATH DBMS DCN OS ALG VB CD SP DM JAVA

DL - 1 1 1 1 1 1 1 1 1 0 0 0 0
C 1 - 1 1 1 1 1 1 1 1 1 1 1 1
OB 1 1 - 1 1 0 0 0 0 0 0 0 0 0
AC 1 1 1 - 1 0 0 0 0 0 0 0 0 0
MATH 1 1 1 1 - 1 1 1 1 1 0 0 0 0
DBMS 1 1 0 0 1 - 1 1 1 1 0 0 0 0
DCN 1 1 0 0 1 1 - 1 1 1 0 0 0 0
OS 1 1 0 0 1 1 1 - 1 1 0 0 0 0
ALG 1 1 0 0 1 1 1 1 - 1 1 1 1 1
VB 1 1 0 0 1 1 1 1 1 - 0 0 0 0
CD 0 1 0 0 0 0 0 0 1 0 - 1 1 1
SP 0 1 0 0 0 0 0 0 1 0 1 - 1 1
DM 0 1 0 0 0 0 0 0 1 0 1 1 - 1
JAVA 0 1 0 0 0 0 0 0 1 0 1 1 1 -

Table 5. Color assignment for Organizational Behavior (OB)

Subjects DL C OB AC MATH DBMS DCN OS ALG

Color 1 2 3 0 0 0 0 0 0

VB CD SP DM JAVA
0 0 0 0 0

Now, we check the fourth row of the course matrix where Ac

is adjacent to DL, C and OB. So, we assign a new color. It is
shown in Table-6.

Table 6. Color assignment of Accounting (AC)

Subjects DL C OB AC MATH DBMS DCN OS ALG

Color 1 2 3 4 0 0 0 0 0

VB CD SP DM JAVA
0 0 0 0 0
Figure 4. Course matrix graph

Now, we check the fifth row of the course matrix, where

Now, we check the second row of the course matrix. Subject
MATH is adjacent to DL, C, OB and AC. So, we assign a
C is adjacent to DL. Thus, we need to assign a different color
to C and we assign color 2. It is shown in Table-4. new color to MATH. It is shown in Table-7.

Table 4. Color assignment of C-programming (C) Table 7. Color assignment of Discrete Maths (MATH)

Subjects DL C OB AC MATH DBMS DCN OS ALG Subjects DL C OB AC MATH DBMS DCN OS ALG
Color 1 2 0 0 0 0 0 0 0 Color 1 2 3 4 5 0 0 0 0

VB CD SP DM JAVA VB CD SP DM JAVA
0 0 0 0 0 0 0 0 0 0

Now, we check the sixth row of the course matrix. Here,

Now, we check the third row of the course matrix where OB DBMS is adjacent to both DL and C, however, it is not
is adjacent to both DL and C. So, we assign a new color to adjacent to OB. So, we could assign DBMS the same color
OB. It is shown in Table-5. which we have previously assigned to OB. It is shown in
Table-8.

© 2018, IJCSE All Rights Reserved 89

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

Table 8. Color assignment of Data-Base Management system (DBMS) Similarly, for VB in row tenth of the course matrix, we need
to assign a new color. It is shown in Table-12.
Subjects DL C OB AC MATH DBMS DCN OS ALG
Color 1 2 3 4 5 3 0 0 0 Table 12. Color assignment of Visual Basic.Net (VB)

VB CD SP DM JAVA Subjects DL C OB AC MATH DBMS DCN OS ALG

0 0 0 0 0 Color 1 2 3 4 5 3 4 6 7

Now, we check the seventh row of the course matrix, where VB CD SP DM JAVA
DCN is adjacent to both DL and C, however, it is not 8 0 0 0 0
adjacent to OB. Here, we need to note that DBMS and DCN
are in the same semester and thus they are adjacent, and Now, we check the eleventh row of the subject matrix. Here,
DBMS has already been assigned the color 3 (which is also CD is a subject offered in the fifth semester and it is not
the color of OB). So, we could not assign DCN the color 3. adjacent to DL. Also none of the subjects offered in the fifth
So, we check the next subject (in the next column) which is semester has the color 1 assigned to it. Thus, we could assign
Ac. DCN and Ac are not adjacent and also the color assigned the color 1, which was previously assigned to DL, to CD. It
to Ac is not yet assigned to any other subjects of the third is shown in Table-13.
semester. Thus, we could assign DCN the same color that we
Table 13. Color assignment of compiler design (CD)
have previously assigned to Ac. It is shown in Table-9.
Subjects DL C OB AC MATH DBMS DCN OS ALG
Table 9. Color assignment of Data Communication (DCN)
Color 1 2 3 4 5 3 4 6 7

Subjects DL C OB AC MATH DBMS DCN OS ALG

Color 1 2 3 4 5 3 4 0 0 VB CD SP DM JAVA
8 1 0 0 0

VB CD SP DM JAVA
0 0 0 0 0 Now, in the twelve row of the course matrix, SP is not
adjacent to DL, but color 1 has already been assigned to CD,
Similarly, we check the eighth row of the course matrix, and CD and SP are offered in the same semester. Thus, they
where OS is adjacent to DL and C, but it is not adjacent to are adjacent and we could not assign the same color to two
OB and Ac. However, color 3 and 4 are already assigned to adjacent subjects. So, we check the next subject which is C.
DBMS-I and DCN-I, which are in the same semester as that C is adjacent to SP. Then we check the next subject which is
of OS. And OS is adjacent to MATH as well. So, we need to OB. OB is not adjacent to SP, and the color 3 is not yet
assign a new color for OS. It is shown in Table-10. assigned to any of the subjects of the fifth semester. So,
assign SP the same color which we have previously assigned
Table 10. Color assignment of Operating system (OS) to OB. It is shown in Table-14.

Subjects DL C OB Ac MATH DBMS DCN OS ALG Table 14. Color assignment of System Programming (SP)
Color 1 2 3 4 5 3 4 6 0
Subjects DL C OB AC MATH DBMS DCN OS ALG
VB CD SP DM JAVA Color 1 2 3 4 5 3 4 6 7
0 0 0 0 0
VB CD SP DM JAVA
Similarly, for Algorithm in row ninth of the course matrix, 8 1 3 0 0
we need to assign a new color. It is shown in Table-11. In the same manner, after observing the thirteenth row of the
course matrix, DM is being assigned the color which was
Table 11. Color assignment of Algorithm (ALG) previously assigned to Ac. It is shown in Table-15.

Subjects DL C OB AC MATH DBMS DCN OS ALG Table 15. Color assignment of Data Mining (DM)
Color 1 2 3 4 5 3 4 6 7
Subjects DL C OB AC MATH DBMS DCN OS ALG
VB CD SP DM JAVA Color 1 2 3 4 5 3 4 6 7
0 0 0 0 0
VB CD SP DM JAVA
8 1 3 4 0

© 2018, IJCSE All Rights Reserved 90

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

Finally, after observing the fourteenth row of the course 2. Set the concurrency value- Concurrency value
matrix, Java is being assigned the color which was determines the number of subjects that could be
previously assigned to MATH. It is shown in Table-16. scheduled for examination in the same time slot.
This value is calculated based on the total number
Table 16. Color assignment of Java (JAVA)
of students registered in all the subjects which have
the same color value assigned, and the total room
Subjects DL C OB AC MATH DBMS DCN OS ALG
Color 1 2 3 4 5 3 4 6 7 capacity (i.e. the capacity of all the rooms taken
together). Suppose, if the total number of students
VB CD SP DM JAVA registered in the subjects having the same color
8 1 3 4 5 value is greater than the total room capacity, then
the concurrency value would be decreased.
C. Sorting the Subjects Based on the Color Value
In the next step, we sort the list of courses based on the color Generating the Exam Time Schedule:
values which are assigned automatically. We have used the We generate the exam time table based on the color value
standard merge sort algorithm [5], as its performance is far assigned to the subjects, the concurrency value, and/or the
better than other sorting algorithms. In worst and best cases number of students registered on each individual subject.
the time complexity is Big Theta ϴ (nlogn). The main
objective of this sorting phase is to reduce the computation Allocation of Rooms:
time for designing the exam time table. Because by 1. We calculate the total number of students registered
performing this phase, the similar types of color assigned in all the subjects which are scheduled in the same
subjects resides together. For this small case study, the time slot.
number of subjects may be less, but in practical scenario 2. We find the room which has the capacity greater
sorting the subjects based on color values will enhance the than or equal to the number of students, and allots
computational performance. that room for conducting the exams for the
So, by considering the same example, after sorting the list of scheduled subjects.
subjects, the subjects are arranged in the following form 3. If the total number of student is greater than any
which given in Table 17. single room capacity, then we allot more than one
room for conducting the exam. And the capacity of
Table 17. Sorted list of courses based on colors. the allotted rooms should be greater than or equal to
the number of students.
Subjects DL CD C OB DBMS SP AC DCN DM
Color 1 1 2 3 3 3 4 4 4 Finally, we print the Exam Time Schedule along with the
rooms allocated. From the implementation point view, we
MATH JAVA OS ALG VB have designed our program using Java, which runs on a
5 5 6 7 8
computer with following configuration: Processor: Intel®
D. Generate the Exam Time Schedule and Allocate Rooms Corei5 2.5 GHz, Operating System: Microsoft® Windows 7
Ultimate, RAM: 3GB.
To generate the final Exam time table, we need the following
inputs:
Continuing with our previous example, let us consider we
have the following amount of data.
1. Fix the number of rooms- First, we need to fix the Say, the total number of room is 3.
number of rooms where the exams would be 1. A55 (Here 55 is the room capacity)
conducted based on the total number of faculties 2. B60
available to conduct the exams. There may be a 3. C70
situation where the number of rooms available
where the exams would be conducted is larger than Total room capacity = 185. The number of students
the number of faculties available to conduct the registered in each subject is given in Table-18. The number
exams. In such a situation, we need to reduce the of faculties available = 7. Now, continuing with our previous
number of rooms where the exams would be held example and the course matrix given in Table-1, along with
such that the available number of faculties could the above details, the exam time schedule generated by the
system would be given below as an Output.
conduct the exams nicely.

© 2018, IJCSE All Rights Reserved 91

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

VI. DISCUSSION
Table 18. Enrolment data of students
The output of the timetabling which our model has generated
Subject No. of Students
for the case study is that the number of days required to
DL 53 conduct the various exams is four (4) and the chromatic
C 55 number is eight (8). For instance the Black color is assigned
OB 50
AC 50
to subject set {DCN, DM, AC}, so according to other
MATH 53 constraint satisfactions, these paper will be conducted on
DBMS 50 same Date and slot. Also, if we observe, particularly a Day
DCN 50
OS 50
say, Fri 01 Dec 2017, in the morning half, Subjects like
ALG 53 Compiler Design (CD) and Digital Logic (DL) are conducted
VB 50 on different rooms‟ viz., Room no: A55, B60. Similarly in
CD 50
SP 50
the afternoon half Subject like C programming is conducted
DM 50 on room no- A55, without any conflicts. So, similarly on
JAVA 50 different Dates also various rooms are assigned properly
without any conflicts. This is very efficient and less time
Output: consuming. Moreover, none of the subjects is overlapped
with any other subjects. So it satisfies the hard and soft
Fri 01 Dec 2017. constraints.
Moreover, for a real time scenario, we have considered
Morning 10:00am to 01:00pm another case study of a University system. From different six
Subject: CD Subject: DL Room: A55, B60 departments, we have collected the subjects name and papers
Afternoon 01:30pm to 04:30pm with their individual code and streams. All total 68 subjects
are there and then we have generated the various Lists of
Subject: C Room: A55
subjects by consulting with the concerned Departments.
Number of Faculty and room availability also collected and
Sat 02 Dec 2017. we have tested in our system. In this case study also none of
the subjects were overlapped with any other subjects. So it
Morning 10:00am to 01:00pm satisfies the hard and soft constraints. The output which we
Subject: DBMS-1 Subject: SP Subject: OB Room: have observed is efficient and satisfactory.
A55, B60, C70
Afternoon 01:30pm to 04:30pm
Subject: DM Subject: DCN Subject: AC Room:
A55, B60, C70

Mon 04 Dec 2017.

Morning 10:00am to 01:00pm

Subject: MATH Subject: JAVA Room: A55, B60
Afternoon 01:30pm to 04:30pm
Subject: OS Room: A55

Tue 05 Dec 2017.

Morning 10:00am to 01:00pm

Subject: ALG Room: A55
Afternoon 01:30pm to 04:30pm Figure5. Final colored Graph
Subject: VB Room: A55
VII. CONCLUSION & FUTURE WORK
The final colored graph generated for the above data is The Time table scheduling problems are directly related to the
shown in Figure.5. number of constraints involved. Higher number of constraints
might raise more numbers of complexities. There is no fixed
algorithm to solve this class of problem. In this work, we

© 2018, IJCSE All Rights Reserved 92

 International Journal of Computer Sciences and Engineering Vol.6(5), May 2018, E-ISSN: 2347-2693

generated the exam time schedule automatically with the help Authors Profile
of graph coloring approach by taking the list of subjects, Rubul Kumar Bania, is an Assistant
number of faculties available to conduct the exams, number Professor in the Department of Computer
Applications, North-Eastern Hill University,
of rooms available for conducting the exams, and the room Tura campus, Meghalaya, India. He has
capacity as inputs. Also, it has successfully satisfied some of completed his Master of Computer
the important hard and soft constraints. The system also Application (MCA) in 2008 and Master of
achieves fairness, accuracy, and optimal exam time period. In Technology in 2012 from Tezpur Central
University, Assam, India. He has published
future, we are planning to add the class routine generation research papers in reputed peer reviewed international journals and
problem with the help of a teacher-subject conflict matrix by IEEE conference and it‟s also available online. He has more than 5
applying edge-coloring with bipartite graph. Also with a large years of teaching experience. His research area of interest is Data
mining, Image processing and Artificial intelligence.
numbers of input data system will be tested.
Pinkey Duarah is a Post Graduate student
and pursuing her Master of Computer
Application (MCA) in the Department of
REFERENCES Computer Application, North-Eastern Hill
University, Tura Campus, Meghalaya, India
[1] Welsh D.J.A., and Powell M.B., “An Upper Bound for the
Chromatic Number of a Graph and it's Application to
Timetabling Problems”, The Computer Journal. (1967),
Vol.10, No.1, pp. 85-86.
[2] Mohammad M., Mohammad A.H., and Osama A.H., “A new
exam scheduling algorithm using graph coloring”, The
International Arab Journal of Information Technology, (2008),
Vol. 5, No.1, pp. 80- 86.
[3] Burke K.E., Mccollum B., Meisels A.,and Petrovic S., “A
Graph-Based Hyper-Heuristic for Educational Timetabling
Problems”, European Journal of Operational Research (2007),
Vol-176, pp. 177-192.
[4] Somasundaram M.R., “Discrete Mathematical structures”, 2nd
edition, PHI, 2010.
[5] Bhasin H., “Algorithms Design and Analysis”, 1st edition,
Oxford University Press, 2015.
[6] Akbulut A., and Yılmaz G., “University Exam Scheduling
System Using Graph Coloring Algorithm and RFID
Technology”, International Journal of Innovation, Management
and Technology, (2013), Vol. 4, No. 1, pp. 66-72.
[7] Hussain B., Basari A.S.H., and Asmai S.A., “Exam
Timetabling Using Graph Colouring Approach”, In the
proceedings of IEEE Conference on Open Systems
(ICOS2011), (2011), pp.139-144.
[8] Verma O.P., Garg. R.,and Bisht V.S., “Optimal Time-Table
Generation by Hybridized Bacterial” Foraging and Genetic
Algorithm”, In Proceedings of International Conference on
Communication Systems and Network Technologies
(CSNT‟12), (2012), pp. 919-923.
[9] Jha. S.K., “Exam Timetabling Problem using Genetic
algorithm”, International Journal of Research in Engineering
and Technology, (2014),Vol.3, No.5, pp. 649-655.
[10] Alon N., “A Note on Graph Colorings and Graph
Polynomials,” Journal of Combinatorial Theory Series B”,
(1997), Vol. 70, No. 1, pp. 197-201.
[11] Gross J. and Yellen J., Handbook of Graph Theory, Discrete
Mathematics and its Applications, CRC Press, Vol. 25, 2003.
[12] Norberciak. M., “Universal Method for Timetable
Construction based on Evolutionary Approach” World
Academy of Science, Engineering and Technology, (2006), pp.
91-96.

© 2018, IJCSE All Rights Reserved 93

View publication stats