Advanced Question Paper Generator using Fuzzy Logic

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 3 | Mar -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 455
ADVANCED QUESTION PAPER GENERATOR USING FUZZY LOGIC
1Pratik Gadge, 2Ravikant Vishwakarma, 3Divya Gandhi
123UG student, Dept of Computer Engineering, Rajiv Gandhi Institute of Technology, Mumbai,
Maharashtra, India
Prof. Suresh Mestry, Dept. of Computer Engineering, Rajiv Gandhi Institute of Technology, Mumbai,
Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The current generation is very competitive in
nature, hence examination is needed at each level which is
used to test the quality of student’s knowledge and expertise
in a particular field. So it becomes necessary to conduct
exams more frequently which is difficult with the traditional
way of manually question paper setting. In this paper we
propose a system named as Advance Question Paper
Generator which provides feature to automatically generate
a question paper. In this system the concept of Fuzzy Logic
in soft computing is used which allows to generate a paper
every time uniquely. The paper can be generated even
before few minutes of examination which denies any
possibility of paper leakage, hence security is also provided.
Key Words: Fuzzy Logic, Analytical, Descriptive,
Fuzzification, Defuzzification
1 INTRODUCTION
Education has become essential thing for human kind.
Education has ability to change the life of a person. It is
very important to evaluate students’ knowledge about
some particular subject he/she is learning, so the system
has to be designed and administered in a systematic and
precise manner. The ongoing examination system i.e.
currently being followed in most of the educational
institutions is very conventional and old. Thus it becomes
inefficient to evaluate the knowledge gained by the
students. The predominant and most followed classical
methodology is that, some predetermined faculties are
responsible for a syllabus and allocated the task of framing
a question paper. In this case, some of the so developed
questions are picked up randomly and used to generate
the paper which evaluates students’ knowledge and thus,
this method is stated as Classical or old-school Method.
This classical method of paper generation suffers from the
following disadvantages: dependency on intelligent people
which might increase the probability of error as they
might have different levels of knowledge on some
particular topic from the syllabus or they might like some
particular questions which results in degrading the quality
of question paper that is being set, there are chances that
proper consideration might not be given to important part
of syllabus ,secrecy may get compromised and full
utilization of resources might not be possible, hence
raising the cost. To overcome these disadvantages,
Advanced Question Paper Generator (AQPG) is being
proposed, so as to make the system more efficient,
reliable, improve the paper setting quality, and also to
reduce the time taken by the instructor in setting the
question papers. It can also help to solve some critical
issues like duplicity, cost, time wastage and above all the
secrecy of question papers. To frame an exam question
paper, various parameters are to be considered; like
difficulty level, amount of numerical type questions,
theoretical questions, weight age of questions according to
marks etc. We cannot exactly tell that this particular
question is difficult and that one is easy. Every person has
different approach to look at the question. Some people
may find it easy, some may find it difficult or some may
find it moderate. Defining exact level of the question like
easy, medium, hard then storing that level in database and
just retrieving those questions from database using simple
queries is not a feasible option. Humans are good in
approximate reasoning. We can have advantage of both
types of reasoning for computation. Fuzzy Logic can utilize
human reasoning effectively. In this paper a Fuzzy logic
based approach is implemented for logical selection of
level of the question, while framing question paper for
every subject irrespective of its discipline. System decides
the paper pattern, difficulty level and what user wants by
taking some inputs from the users. Also system can extract
data from various previously existing databases which
also reduces the work of manually entering questions into
the database. Use of random function will give new paper
every time, However probability of giving same paper two
or more times is almost negligible.

2 LITERATURE SURVEY
We have done the survey of existing papers and found
following information:
In the first paper we referred fuzzy logic and apriori
algorithm is used for generating question paper. In this
login module can differentiate between admin and staff.
Admin fills the information about staff members, give
them access credentials and various rights. All the
operations such as question insertion, modification,
deletion are possible in the system. System uses ‘TOP N
Query’ method of fuzzy logic. This method limits the
number of rows returned from ordered sets of data. Also
TOP N query method chooses random questions every
time. Apriori algorithm is used for data mining where
importance is given to the frequently occurring items.
Paper is generated according to user defined level. Paper
is stored in txt format. In Apriori large number of
questions sets are analyzed every time and hence it results
in large number of computations. [1]
In the second paper, they have used fuzzy logic to
determine the level of the questions. In this all users for
particular question enter analytical and descriptive
indices. They have divided this process into three stages.
In this system, analytical and descriptive parameters may
alter depending on all users’ response. Changing values
increases number of calculations and hence workload
increases. [2]
In this System question paper containing multiple choice
questions is generated. System is based on two techniques
which are fuzzy logic and ants’ colony. In this system
admin has rights to enter the questions and their
respective options into the database. Admin also provides
the weightage and complexity level for the questions. In
this admin chooses type of the paper (which can be easy,
medium or hard) and generates question paper. After
generation of pdf file, System makes email to the
appropriate college. It is mainly required for competitive
exams. [3]
3 EXISTING SYSTEM
The current system is manual and used in most of the
institutes and universities which says that one or two
person has authority to set the paper. Some of the
professors just refer previous year paper and generate
paper with very small amount of innovative work.
However in this there is major problem of security. It may
cause paper leak. This conventional system does not
consider all respective professors’ view at the time of
question paper generation. Also the time taken by this
method is also more. Other System we referred is based on
apriori algorithm which gives importance to frequent item
sets due to which some questions may omit totally and
won’t be included at all. This may also omit some
important questions.
4 PROPOSED SYSTEM
Our AQPG is based on fuzzy model which will be
implemented using Microsoft visual studio. We will use
Microsoft SQL server as a database management system.
Our System will have registration and login functionalities.
After registration, the credentials provided by the user will
be verified by the admin.
If user is authenticate then admin will make his username
and password valid and access will be granted. While
registration user can select maximum four subjects for
which he wants to set the paper. Admin has rights to
change the subjects of a particular user. Each question has
attributes like analytical index, descriptive index and
marks. While entering the question into database,
registered user has to enter its analytical, descriptive
indices and marks. Analytical index specifies amount of
thinking or reasoning required to solve a particular
question. Highly analytical index means question tends to
difficult level. Descriptive index means how much
theoretical the question is. User can also modify Analytical
index, Descriptive index of the question. Average of
previously existing indices and user entered indices will
be stored into database. User can also delete the question.
If all users of that respective subject are agree for deletion
then that question will be deleted. These indices are fuzzy
input descriptors. Fuzzy output descriptor for a particular
question is the level of the question. We have five levels
namely-very easy, easy, medium, difficult and very difficult.
We will make the system distributed with questions’
database deployed on one node or server.
5 METHODOLOGY: FUZZY LOGIC
Lotfi Zadeh, a professor in California is founder of fuzzy
logic. The use of fuzzy logic has increased considerably in
wide variety of applications of fuzzy in recent years. The
applications comprises of hardware as well as software
products. We know that humans are very good in
approximate reasoning and can interpret vague
information. We have two types of sets – fuzzy and crisp.
Crisp set theory says that degree of membership of an
element to a particular set can be either zero or one.
Consider example in which we have two sets even and odd
with natural numbers as a universal set. In this example,
number can be either even or odd. There is fixed boundary
between even and odd numbers. So there is no confusion

and we can have 100% degree of membership for any
number to a particular set. But some real life estimations
cannot be derived using crisp set theory. Consider height
of a man, say 1.8 meter. We cannot exactly tell whether he
is tall or medium heighted, but we can have degree of
membership to both sets (like 0.7 for ‘medium’ and 0.4 for
‘tall’). This degree also varies from person to person. Some
people may say that 1.8 meter height falls in tall category
while some may say that it falls in medium category.
Question, marks, Analytical, descriptive indices
Consequent
For (‘then ‘part)
Matched rules
W IF then Rules
X y Z
D L
W=Analytical & descriptive indices (crisp form).
X= Analytical & descriptive indices.
Y=sets corresponding to indices & their membership values
for those sets.
Z=set of matched rules, membership values.
D=Question and its level.
L= Level of the Question
Where X, Y, Z are labels for the arrows which are
above them.
Important components of fuzzy model with
respect to our system
5.1 Questions’ database: This is the database which
contains various subjects’ relations (tables). Attributes of
subject relation are serial number, question, Analytical
index, Descriptive index, marks.
5.2 Fetch stage: This is a block of code which fetches
the questions after getting particular subject’s name from
a valid user. It passes Analytical index and Descriptive
index to the fuzzification block.
5.3 Fuzzification: After getting crisp input from fetch
stage,
This block decides appropriate fuzzy sets for these crisp
values and gives membership values for those sets. For
example: Particular question’s membership value (u) for
‘difficult’ set is 0.8 and ‘very difficult’ set is 0.5. Where
0<=u<=1.
5.4 Fuzzy Rule base: This module stores fuzzy rules
generally in the form of two dimensional array. This
module is read-only and is designed by the system
developer. However storage technique or data structure
for rule base is also decided by the system developer. This
module should be designed carefully because it will play
very important role while decision making.
5.5 Inference engine: This is also very important
module. This will match if then rules and select some rules
out of predefined rules (say 4 rules out of 16).
Membership values for those rules will be passed.
5.6 Defuzzification: This block uses min-max
composition function for our system. It takes membership
values from previous stage, does certain computations and
finally gives level of the question.
5.7 Decision Maker: This module gets level of the
particular question from defuzzification stage and user
selected level. If user selected level matches with the level
given by Defuzzification block, then question is included
otherwise rejected.
6 CONCLUSIONS
With this paper we present a system which is standalone
application as Advance Question Paper generator designed
for educational organization in such a way that, a question
paper is generated with the set of predefined questions in
the database. Operations like question insertion,
modification, deletion are also possible in this system
which makes it dynamic. In this, it is ensured that each
time the question paper generated is unique. With this
system the question paper can be generated on the day of
examination, so it eliminates the chance of paper leakage.
ACKNOWLEDGEMENT
I would like to express special thanks to our professor Mr.
Suresh Mestry and H.O.D of computer engineering
department Prof. S.Y ket for their help, guidance and
support.
Questi
ons’ DB
Fetch
stage
Fuzzif-
ication
Fuzzy
rule base
Register
ed User
Defuzzi-
fication
Inference
engine
Decision maker

REFERENCES
[1] Suraj Kamya, Madhuri Sachdeva, Navdeep Dhaliwal
and Sonit Singh, “Automated Question Paper Generator
System using Apriori Algorithm and Fuzzy logic”,
International Journal for Innovative Research in Science &
Technology, (IJIRST), 2016.
[2] S. Kamya, M. Sachdeva, N. Dhaliwal, and S. Singh,
“Fuzzy logic based intelligent question paper generator”,
IEEE Advance Computing Conference (IACC), 2014.
[3] Pushkar Ambole, Urvashi Sharma, Pushkar Deole
"Intelligent Question Paper Generation System",
International Journal of Scientific and Technical
Advancements(IJSTA)
[4] S.N.Sivanandam, S.N.Deepa "Principles of Soft
Computing" Second Edition, Wiley Publication

Advanced Question Paper Generator using Fuzzy Logic

More Related Content

What's hot (19)

Similar to Advanced Question Paper Generator using Fuzzy Logic (20)

More from IRJET Journal (20)

Recently uploaded (20)

Advanced Question Paper Generator using Fuzzy Logic