Volume 3, Issue 3, March-2016, pp.

112-118 ISSN (O): 2349-7084

International Journal of Computer Engineering In Research Trends

Available online at: www.ijcert.org

Course Scheduling in University using Agent

Exploration in Timetable Media
Jamaludin Hakima,1, Sri Hartatib,2
a
YAPIS Papua University, Jl. Sam Ratulangi Dok V Yapis,Jayapura 99111, Indonesia
b
Gadjah Mada University, Address,Yogyakarta, Indonesia
1
j2hakim@yahoo.com; 2shartati@ugm.ac.id

AbstractCourse scheduling in University is still a study area that is interesting for researchers to explore and search its optimization
[4]. Scheduling optimization still depends on the needs of a University. Manual scheduling is still the most commonly used option, but it
requires a skilled and expert person or person who has been accustomed to do so. However, the result still needs a lot of corrections.
Course scheduling in College/University in Indonesia using semester credit system (SKS) is a challenge in making course sched ule. Some
scheduling methods have been proposed in several studies such as genetic methods [10], integer programming [1] and other methods
using other agents [2] [7]. In the previous studies, the generated schedules must still be adjusted and or inflexible when th ere are changes
in the schedule or its pattern, and its model cannot be applied to all Colleges/Universities in Indonesia using semester credit system (SKS).
This study offers a method which is referring to the constrains/conditions in Universities but still flexible in its formation and change. Agent
will explore timetable media to locate time cell/slot corresponding to the scheduling using credit system (SKS).
Index TermsAgent Exploring, Schedule, Timetable Media, Rules, Conditions

1. INTRODUCTION Timetable media exploration method will make
course scheduling manually. All conditions used in
Course scheduling is still an interesting thing for
making schedule will be agent behaviors, thus, it is
researcher to study its optimization [4]. The complexity
expected that problems in remembering conditions that
of course scheduling depends on the number of
constrains/conditions to be met. Each College has its are carried out manually can be overcome. The results
own conditions/rules in making schedule. Such using media exploration schedule by agent will be close
rules/conditions will make the method for making or equal to the optimization conducted manually.
schedule slightly different. Those rules/conditions will
change into constrains/rules in making course 2. RELATED WORK
schedules.
There have been many studies conducted by
Schedule resources such as room and timing will be researchers to identify the optimization in course
things that affect the making of course schedule to scheduling. Various methods have been used to
comply with the conditions. The lack of rooms to be formulate the course scheduling. Genetic Algorithm
used in the course schedule will make the course (GA) is one of the most popular methods in finding an
schedule more complex. Similarly, if available timing in optimization, including for course scheduling. One of
the form of everyday time shift/slot is inadequate, it techniques using GA is by changing course schedule
will cause that not all conditions can be met. components into chromosome, and then, sub-
chromosome, and finally, mod-chromosome [10].
Manually course scheduling is still regarded as the Integer Programming is also one of methods that
most optimal method, thus, many convert course schedule components into numbers 1
Colleges/Universities still apply it; or the results of and 0, in order to obtain the optimal combination in the
automatic method will be adjusted manually. Manual course scheduling [1]. However, a mathematical
scheduling requires personnel having the required method mentioned earlier is used to obtain the most
expertise or persons who have done it for many times optimal combination in order to place the course
and know how to make course schedule in each schedule component according to the applied
semester. Considering that all conditions that must be conditions. These methods are effective in generating
applied to the course scheduling should be done for course schedule but some adjustments are still needed
each schedule component to be scheduled, the problem and it will inflexible when there is a change in the
will be the accuracy when it is done manually.
2016, IJCERT All Rights Reserved Page | 112
Jamaludin Hakim et al., International Journal of Computer Engineering In Research Trends
Volume 3, Issue 3, March-2016, pp. 112-118

course schedules. In case of any adjustments or changes used. If the hard constrain cannot be met then the
in the course schedule, all schedule components will be schedule cannot be used. Hard constrain usually
rearranged, thus, the unchanged or unmoved schedule consists of three items:
components will still be affected by these changes.
a. A lecturer cannot be scheduled more than once
Agent has been used in some studies related to at the same time period
scheduling. In the scheduling for transportation, agent
b. A classroom cannot be scheduled more than
will act as operator confirming the passenger pick-up
and carrying in taxi scheduling in real time scheduling one course at the same time period
[3]. From these studies, agent is assigned to replace the c. Students at same level/semester cannot be
role of taxi operator. On the activities in factories, scheduled at the same time period.
agents can also be assigned to manage resources to be
used[8]. Agent has replaced the human roles but it did Sometimes, at the time of making course schedule, a
not has ability as human in regulating. For the case of College/University has rules or regulations other than
room booking in hospital, agent will replace the role of the hard constrains that govern lecturers, use of
receptionist to arrange the booking order and the use of classroom, and the students in the course scheduling.
rooms in surgery and post-surgery process [6]. The generated schedule will meet the users desire,
which means that it is optimal from the perspective of
In the same case, agent is used for exam scheduling. the users if it can comply with the conditions from the
Schedule is arranged for students in a university to be University. These conditions are called as soft constrains.
present in different semester exams. Schedule shall be Some examples of soft constrains that are generally
set in such a way so that students will not have more exist in College/Universities in Indonesia are:
than one exam at the same time [7]. This technique does
not use time slot referring to the semester credit system a. Lecturers are not scheduled consecutively in one
(SKS) SKS in which each SKS is allocated with 50 day, there is a lag time between one course with
minutes in classroom, on the other hand, the use of the other, in order to give opportunity to break or
Utility-Base-Agent-System is less optimal in many prepare materials for the next class.
spread time slots. In Tambuk University, multi-agent b. A room can only be scheduled twice
system is used to form a pattern of course type to be consecutively, to give opportunity for arranging
scheduled [2]. This technique has not been able to or cleaning the room.
accommodate course scheduling with SKS models, in
addition, this modeling cannot be applied to the c. Students at the same level/semester should not be
courses in Indonesia. scheduled more than three times in one day

3. AGENT EXPLORATION METHOD d. Elderly lecturers or lectures having health

problems shall not be scheduled on rooms located
Colleges/Universities in Indonesia using credit on the top floor
system (SKS) as its learning model requires courses
scheduling in accordance with that system. Each e. The scheduled room should be within its capacity;
subject has a weight which is also an indicator for the otherwise, the room capacity must be greater than
amount of time required in teaching in classroom. the number of participants.
According to regulations, 1 credit is equal to 50 minutes For the conditions of choice, not all universities are
in the classroom, so that if a subject has 3 credits, it same, but there are some conditions that are often used
takes 3x50 minutes = 150 minutes per week. by universities based on the research result and health,
Manually, schedule draft will be formed in row and such as providing opportunity for lectures to take
column model that represents day, time slot (1 slot = 1 enough break (not lag time for class movement) to
credit = 50 minutes) and classroom. Schedule draft will refresh their mind and energy and or to prepare
be filled according to the available time and rooms materials and or equipments. Conditions of choice are
based on course weight/credits, which means that not an absolute requirement to be met because it will
course having 3 credits will get 3 time slots on the depend on resources being owned and how the
schedule draft. To perform these activities, personnel College/University overcomes these problems.
making schedule will manually conduct visual On manual scheduling, personnel will conduct a
exploration to find a time slot corresponding to subjects visual exploration on schedule draft to find the suitable
and meet the conditions/constrains. time slot [4]. In the proposed method, agent will
3.1. Conditions/Constraints perform the task to explore the schedule draft and to
find a time slot suitable with the applied conditions.
Indicators for the successful of making course schedule Schedule draft becomes a very important media in
is how to meet the conditions, particularly the hard using this method, thus, the schedule draft will be the
constrains. In scheduling, there are two types of media used by agent to conduct exploration,
constrains, the first is hard constrain and the second hereinafter referred to as "Timetable media"
one is soft constrain [7]. Hard constrain is mandatory
conditions that must be met so that a schedule can be

2016, IJCERT All Rights Reserved Page | 113

Jamaludin Hakim et al., International Journal of Computer Engineering In Research Trends
Volume 3, Issue 3, March-2016, pp. 112-118

3.2. Timetable Media

The timetable media will be like a chessboard in a chess
game, while agent will be like the pawn that move from
one field to another based on the rules. The timetable
media will consist of rows and columns. The rows will
contain information about the time slots (day and time
shift) and will contain information about rooms. The
visualization of timetable media is seen in Figure 1.
Each cell in timetable media represents a timing
unit in Semester Credit System/SKS (50 minutes). The
amount of shift time on the media schedule depends on
each college/ university, but the average is nine or ten
shifts which spread from morning until afternoon. If Fig. 2. Structure of Timetable media
colleges/universities do course until night then the
schedule shift will be more and there are more cells that 3.3. Agent Exploration Model
can be used.
Exploration model consists of four components. The
Referring to the visualization of timetable media in first is user who will enter the schedule components
which the information will be saved and will generate a (course code, lecturer code, course level/semester,
schedule, the structure of timetable media table study program and number of participants). The
second one agents that will explore timetable media by
structure will contain information mentioned themselves; the third is timetable media that will
previously. In addition to information about slot time contain information about the results of agent
and classroom, timetable media will also contain exploration; and the last is the course schedule. Agent
exploration model can be seen in Figure 3
information about course, lecturers, course credit,
semester/level, classroom capacity, study program and In agent exploration model, user will enter the
the room floor. The structure of timetable media can be schedule components in which the information will be
used by agents to trace cell by cell in timetable media to
seen in Figure 2 find a time slot in accordance with the schedule and
following the conditions. All schedule components will
firstly be kept in the storage space. The schedule
components will be taken by agents in accordance with
the order being stored.

Fig. 1. The visualization of timetable media

Fig. 3. Agent exploration Model

Agent will check cell by cell based on the conditions

specified as agent behavior. Agent checking will be

2016, IJCERT All Rights Reserved Page | 114

Jamaludin Hakim et al., International Journal of Computer Engineering In Research Trends
Volume 3, Issue 3, March-2016, pp. 112-118

carried out section by section from the media schedule. 5. Checking whether any lecturers who teach two
Agent does not directly check all parts of its timetable times respectively, if lecturers are found to be
media/environment. Section by section check if for scheduled at a position before the agent cell for all
efficiency in the condition checking. Based on agent rooms, then the agent will move to other shift
checking activities carried out to some cells of timetable vertically (h,s++, r), if the lecturers cell is found after
media, the type of agent used is Simple-Reflex-Agent the current agent position as much as total
[9]. Agent only reacts to the conditions faced at that weight/credits for all rooms, then the agent will
time and takes action in accordance with the move shift as much weight/credits course plus one
rules/conditions. (h, s+credits/bbt +1, r), if it is already in its final
position shift, it will move to other day (d++, s, r)
3.4. gent Exploration Algorithm
Agent exploration will be carried out from cell to cell in
timetable media starting from the initial cell. Cells on
media schedule will be marked with day, shift and
room (d,s,r). The initial cell will be marked with 1,1,1
which means that the course is on cell of first day
(Monday), first shift, and first room. The movement of
agent in exploring timetable media is determined by
the conditions and provisions that must be executed.
Room/ horizontal movement will add room value (d, s,
r ++), shift/vertical movement will add shift value (d, s
++, r), and day/vertical movement will add day value
(d++, s, r).
Course will be placed by agent on the timetable
media based on credits of the course. Course having
three credits will occupy three shifts vertically, as well
as courses having two credits. The course placement in
the cells of timetable media will refer to the course
placement algorithm that are adjusted with the
conditions. The course placement flow is depicted in
Figure 4.
Agent will execute condition by conditions owned
by agent if it is in the particular cell position. Each
constrain will experience with a condition of being met
or not. If being met, agent will check the next
conditions. If it is not met then the agent will move to
other cell in accordance with the existing condition,
whether it only moves to other room, switch to other
room or move to other day. Here are the criteria for
condition and agent movement:
Fig. 4. Agent Exploration Flow
1. Checking room capacity, if the room capacity does
not match, then, it will be moved to the next room
(d, s, r++) Based on constrains and agent movement, the
2. Checking the possible/empty rooms, if following methods can be formed:
impossible/not empty, then it will move to other Agent movement
room (d,s,r++) MoveRoom(room);
3. Checking lectures who teach for all rooms as much {
as the weight/credits, if a lecturer was found in one ifroom=maxroom thenroom=1;
room for all shifts as much as weight/credits, then roomroom++;
the agent will move to other shift/cell vertically as return room;
much as weight/credits of the current agent }
position (d,s+credits/bbt,r).
4. Checking student semester/level, if the same MoveShift(shift, credit);
student semester/level is found in all rooms as {
much as the course weight/credits, then agent will localday;
move to the next shift vertically as much as course ifshift=maxshiftthen
weight/credits (d, s+credits/bbt, r) {
shift=1;

2016, IJCERT All Rights Reserved Page | 115

Jamaludin Hakim et al., International Journal of Computer Engineering In Research Trends
Volume 3, Issue 3, March-2016, pp. 112-118

MoveDay(day); returnlecresult;
} }
Shiftshift+credit;
returnshift; CourseLevel(day, shift, room, sublev)
} as boolean;
{
MoveDay(day); localsublevresultasboolean;
{ local sublevstatasstring;
ifday=maxdaythenday=1;
dayday++; sublevstat search for course
returnday; level/semester in cell starting from
} (day, shift, room) until the end room
(roomax) as much as course
Constrains weight/credits (shift++ as much
weight/credits);
ifsublevstat =
RoomCapacity(day, shift, room,cap) sublevthensublevresultfalse
asboolean; elsesublevresulttrue;
{ returnsublevresult;
localroomcapasstring; }
localcapresultas boolean;
LecTeach(day, shift, room, idlec)
roomcaptaking room capacity in asboolean;
position (day, shift, room); {
ifroomcap=capthencapresulttruee Locallectearesultasboolean;
lsecapresultfalse; locallecteastatasstring;
returncapresult;
} Lecteastatsearch for lecture
code at the beginning of shift before
RoomAvailable(day, shift, room) the current shift (day, shift--, room)
asboolean; until the end room and return to the
{ room before this shift;
localroomresultas boolean; iflecteastat =
localroomstatas string; idlecthenlecstearesultfalse
else
roomstattaking room volume in {
position (day, shift, room); Lecteastat = empty;
ifroomstat = Lecteastat searching for
emptythenroomresult false lecture code in shift position after
elseroomresulttrue; the current shift is added with weight
returnroomresult; (shif+weight+1), until the end room and
} return to the room before this shift;
iflecteastat =
idlecthenlectearesultfalse
LecturerAvailable(day, shift, room, elselecteastattrue;
idlec) asboolean; }
{ returnlectearesult;
locallecresultasboolean; }
locallecstatasstring;

Lecstat search for lecturer 4. RESULT AND DISCUSSION

code in cell starting from (day, shift, Methods that have been made will be encoded using
room) until the end room (roomax) as JADE (Java Agent Development Environment)
much as course weight/credits (shift++ Framework. Neatbean Editor will help in making the
as much weight/credits); program. JAVA programming will be the basis in
iflecstat = making exploration agent, but the object class will use
idlecthenlecresultfalse JADE library. In JADE Framework, agent will be
elselecresultrue; created in a container, and each container can consist of

2016, IJCERT All Rights Reserved Page | 116

Jamaludin Hakim et al., International Journal of Computer Engineering In Research Trends
Volume 3, Issue 3, March-2016, pp. 112-118

several agents, but in this paper, the container only

consists one single agent.
The conditions in which the methods have been
made will be entered into object class of agent behavior.
Each condition will be made separately into methods in
JAVA. These conditions will be invoked in agent
behavior class. In agent behavior class, a central
program will be built to sort and test each condition
methods. Each condition method will be run
continuously as long as it generates false value and will
continue to the next method if it has true value. If all
method conditions have been fulfilled, then, the agent
will stop by itself. It means that all value components
have occupied cells on the timetable media by meeting
the existing conditions. The quotation of the program
code of agent behavior is shown in Figure 5.

Fig. 4. Agent Exploration Flow

The entire result of schedule test could not be

displayed because of the resulted schedule display is
quite long. In the display of the resulted schedule, it
can be seen that schedule components consist of course
code, lecturer code, semesters, study program code and
course weight/credit. The display is still in the form of
code in order to observe whether the placement of
schedule is conforming to the stipulated conditions.
From these results it appears that certain courses will
occupy cells as many as the course weights/credits
being owned, similarly, a lecturer will not be scheduled
to teach at the same time, so did students of the same
level/semester in the same study program will not be
scheduled at the same time.
Fig. 5. Quotation of agent behavior code
One of advantages of this agent exploration method
is in the schedule change or adjustment. In that process,
In this paper, an experiment has been done using agent the schedule will not undergo major changes given that
exploration to make a course schedule by using 10 most of schedule components have already obtain the
classrooms, six days and 9 shift per day and 20 courses, right timing and room. The step for changing schedule
7 lecturers as the schedule components to be scheduled. will be the same with the process for making schedule
The time required for executing 5 condition is 40 in which only component values to be changed will be
excluded from timetable media by recording the
seconds. The time required will vary depending on the
schedule component order to be scheduled. Ordering position that has been occupied, so that at the time of
schedule components can either accelerate or decelerate exploration, it will search for the time slot and the
the scheduling process. appropriate room, a position that has been recorded
will be ignored.
In this study, the time required to make schedule is not
an indicator for the success or optimization of 5. FUTURE WORK
scheduling. The accommodated rules applied to a This study is still very far from complete, even, it can
schedule will be an indicator for the success or be said as a preliminary study. This method is also
optimization of scheduling using agent exploration highly dependent on the component ordering to be
method. The more conditions to be applied, the more scheduled. Ordering using certain methods such as
complex the agent exploration process and the more larger weight/sks or the lectures being scheduled in
time needed. advance can be additional research.
Ignore a condition of choice will be one of alternatives 6. REFERENCES
to make agent exploration not repeated because it does
not meet its condition. The use of schedule component [1] Bakir, M.A and Askop, C, A 1-0 Integer
order to be scheduled may also be an option. The Programming Approach to a
results of agent exploration experiment can be seen in UniversityTimetabling Problem, Hecettepe
Figure 6. Juornal of Mathematics and Statistics, Vol 37(1), pp.
41-55, 2008.

2016, IJCERT All Rights Reserved Page | 117

Jamaludin Hakim et al., International Journal of Computer Engineering In Research Trends
Volume 3, Issue 3, March-2016, pp. 112-118

[2] El-Helly, M., Abdelhamid, Y. dan Al-Wakeel,

M.,A Multi-Agent Pattern Based Timetabling
System, EgyptianComputer Science Journal,Vol.
35, No. 3 September 2011
[3] Glascheko, A., Ivaschenko, A., Rzevski, G. dan
Skobelev, P., Multi-Agent Real Time Scheduling
for Taxi Companies, 8th International Conference
on Autonomous Agents and Multiagent
Systems,10-15 may 2009, Budapest, Hungaria.
[4] Hakim, J and Wardoyo, R.,Using Grid for Course
Scheduling with Multi Agent, International
seminar of Information Technologi (ISIT), October
23th, 2010, Bandung, pp 147-149.
[5] Hosny, M. and Fatimah,S.,A Survey of Genetic
Algoritms for University Timetabling Problem,
International Conference on Future Information
Technology,IPCSIT vol. 13 (2011) IACSIT Press,
Singapore
[6] Hutzschenreuter, A. K., Bosman, P. A. N. dan
Blonk-Altena, I.,Agent-based Anmission
scheduling in Hospital, 7th International
Conference on Autonomous and Multiagent
System (AAMS 2008),pp 45-52, Estoril, Portugal
[7] Mehmood, T. dan Farid, M. S.A Multi-agent
Architecture for Task Scheduling In University
Environment, IAES International Juornal of
Artificial Intelligent,Vol. 1, No. 4, Des 2012, pp.
193-200, ISSN:2252-8938.
[8] Paulo, R. dan Rocco, P., Adaptive Dynamic
Scheduling in Agent Based Manufacturing
Environment : A Budget Approach,
Communication of IBIMA, Volume 7, 2009 ISSN :
1943-7765
[9] Russell, S. and Norvig, P.,Artificial Intelligence: A
Modern Approach, 2nd ed. Prentice Hall. 2003,
New Jersey, USA.
[10] Suryadi, D. and Philipus, R.Genetic Algorithm For
University Timetable Planning in FTI, Proceeding
of the 2012 International Conference and Industrial
Engineering Operation Management, , July 3-6,
2012, Istambul, Turkey

2016, IJCERT All Rights Reserved Page | 118