Vikalp - Automatic multiple choice questions generator

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 3207
Vikalp - Automatic multiple choice questions generator
Amit Singh1, Parth Kalbag2, Divya Kukreja3, Ritu Kalyani4
1Assistant Professor, amit.singh@ves.ac.in, Dept. of AI and Data, Vivekanand Education Society’s institute
of Technology, Mumbai, Maharashtra, India
2-4Student, 2018.parth.kalbag@ves.ac.in, 2018.divya.kukreja@ves.ac.in, 2018.ritu.kalyani@ves.ac.in, Dept. of
Information Technology, Vivekanand Education Society’s institute of Technology, Mumbai,
Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In any education system, examinations are
conducted to judge the caliber of the students. To conduct the
examination, teachers need to generate the questions
manually which is a very time-consuming process. To reduce
time and effort, a system through which multiple choice
questions can be automaticallygeneratedforuser-giventextis
proposed in this paper. Fill In the Blanks, True or False and
Match the Following are the typesof multiple-choicequestions
covered.
Key Words: Automatic Multiple choice questions,
Natural Language Processing, Mcq’s, Distractors,
Conceptnet, Wordnet, Sense2Vec.
1. INTRODUCTION
In this modern world where technology is always evolving
and has increased its reach to various sectors of the
educational industry, various schools and colleges have
adopted the e-learning platforms in which they assess
students through various exams like GATE, CAT, College
Exams, etc. it has be-come quite a cumbersome task to
generate multiple-choice questions for the whole syllabus.
MCQ-style questions are a basic instructional tool that may
be used for several reasons. These types of questions can
affect student learning in addition to serving as an
assessment tool. In certain examinations, the only way of
ensuring proper assessment of students is by generating
good quality MCQs so that they can explore much more than
usual and are well-suitedtothecurrent e-learningmeasures.
The paper focuses on generating various types of MCQs like
Fill in the Blanks, True or False, and Match the following.
Multiple Choice questions are a form of evaluation in which
respondents are inquired to choose the most appropriate
reply out of the choices from a list. The MCQs are formed
from basically two entities: a question and various possible
options including the correct answer.The questionisformed
by determining which type of question can be formed. For
e.g., “The number of days in a week are ” For the following
sentence as the sentence contains a numerical value so the
type of question generated can be a numerical one with the
correct answer replaced by a “ ” and suppose considering
another example “Still, numbers for server use of Windows
(that are comparable to competitors)showonethirdmarket
share, similar to that for end user use.”. If this is the
respective question generated, then the type of MCQ the
system should generate is either a True or False or a Fill in
the blank type of MCQ.
The major challenge in generating any MCQ is the
requirement for great distractors, i.e., distractors ought to
show up as a conceivable reply to the question indeed to an
understudy with great information on thespace.Atthesame
time, it should not be a substitute reply orsynonym.Besides,
a well written MCQ should contain sufficient data to answer
the question. For example: “The color of the blood is .” and
the corresponding distractors are: Red, Maroon, Dark Red,
Crimson Red. As in this example all the distractorshavevery
similar meaning. Hence,the distractorsshouldactuallymake
more sense as in this example: “The color of the blood is .”
and the corresponding distractors are: Red, Blue, Green,
White. Here, all the distractors have similar context but are
clearly distinguishable from each other and are making a
good impact.
2. LITERATURE SURVEY
This section discusses research works which proposed the
idea for Automatic MCQ generation. Cloze Questions contain
questions where questions contain one or more blanks and
multiple choices listed to pick an answer isdiscussedin[1].It
was actually a goal-oriented system, that is, a specific field
based on cricket world cup data. Cloze system is divided into
three modules: sentence selection, keyword selection and
distractor selection. So, the end output gives an English
article on cricket World cup and the system generates Cloze
questions.
Real time multiple-choice question generation for language
testing and English grammar is discussed in [2]. For the
application, NLP technique and basic machine learning semi
supervised algorithm such as Naive Bayes Classifier and
KNearest Neighbors algorithmwereused.Areal-timesystem
generates only one type i.eFill in theBlankstypeofquestions
on English grammar and vocabulary from online news
articles which takes an HTML file as input and turns it into
the quiz session.

The next research paper that was discussed and examined
was based on generating MCQ questions using string
similarity measure in [3].The research paper mainly focused
on keyword selectionandgeneratingdistractorsbasedonthe
1 semantic labels and named entities in text and string
similarity measures respectively. So, for selection of
sentences and keywords, it is based on the semantic labels
and named entities that exist in the sentence. And for the
distractor generator concept of similarity measure between
the sentences is used. Inthisproposal,threealgorithmsofthe
character based type and five algorithms of the term based
type to measure the similarity between two sentences were
used.
Similarly, another system is proposed which generates the
automatic MCQ based questionsusingthetextextractedfrom
the web in [4]. So mainly web scraping is done, and it
summarizes the text using the technique of fireflies
preference learning. So, the main part is sentences and
distractors. Theytransformedthesentenceintostemsandfor
distractor generation they usedthesimilaritymetricssuchas
hyponyms and hypernyms. Also, the system is used to
generate the analogy questions to test the verbal ability of
students.
3. PROPOSED METHODOLOGY
The system takes the input text from the user and generates
MCQs. The MCQs that will be generated are inthemixedform
of Fill in the Blanks, True or False and Match the following.
The system would analyze the text by summarizing the text
using the “BERT - State of Art” language model algorithm.
Besides, all the watchwords from the summarized content
are extracted using POS tagging. At that point the Sentence
Mapping is done by extracting the sentences for each
keyword. Within the Sentence Mapping, all the sentences
from which MCQs are to be shaped are extracted. Since the
system isn’t fairly kept to one sort of MCQ, thus classification
is an important step of the system. All the sentences which
were extracted within the Sentence Mapping phase are
classified here as which sort of MCQ should be shaped from
which sentence. And after that based on the sort of sentence
classified,comparingDistractorsarecreated.Atlast,different
techniques are used so that the user would get syntactically
correct MCQs. The stepscarried out by the systemareshown
in Fig -1.
Fig -1: Proposed System
A. Summarization
For the text summarization the system is using a predefined
BERT (Bidirectional transformer) Extractive Summarizer
model. It is a state of art language model in which minimum
length and maximum length is provided to the model to
generate summarized text.
B. Keyword Extraction
After finding the relevant information from the generated
summarized text the next step is to find all the important
keywords which can be treated as an answer to MCQ. Hence,
POS tagging is used so that all the relevant nouns, proper
nouns,adjectives, and verbs are extractedaskeywords.Then
they are sorted in descending order according to the
preference of the best ones.
C. Keyword Sentence Mapping
Keyword sentence mapping is done by tokenizing the
sentences. Here, all the sentences of the corresponding
keywords are extracted and tokenized. These are the
sentences from which corresponding Multiple choice
questions will be generated.
D. Classification
If the sentence contains any numeric value, then those
sentences are classified into numeric MCQs category as such
type of MCQs would be more accurate,elsearandomnumber

is generated between 0 to 1 and if the numberisbetween0to
0.75 then Fill in the Blanks MCQ would be generated and if
the number is between 0.76 to 1 then True or False type of
MCQ would be generated.
E. Generating Distractors
A good distractor is one that gives similar meaning as the
answer but is not the answer itself. The goodness and
toughness of an MCQ question depend on how close the 2
distractors are to the key. The closerthedistractorsaretothe
key, the more difficult the question is. This step is
independent for each type of MCQ. Based on the category of
the MCQ that is to be formed from the sentence,
corresponding types of distractors are generated. Before the
generation of distractors, ‘word sense’ is first applied for the
answer and based on that sense the distractors are
generated.
4. ALGORITHMS
This section discusses the implementation part of our
proposedsystem.Firstweshowtheoutcomeofcoep-package
developed to standardise the question generation process.
Next, we show the integration of the proposed system with a
website through APIs along with the Performance Analysis
and Navigation part.
A. Fill in the Blanks
For the text summarization the system is using a predefined
BERT (Bidirectional transformer) Extractive Summarizer
model. It is a state of art language model in which minimum
length and maximum length is provided to the model to
generate summarized text.
For the Fill in the blank type of MCQs, the distractors are
generated by 2 ways. They are as follows:
1) Wordnet: For all the dictionary keywords wordnet works
well. It is used to determine the similarity between words.
The Wordnet algorithm measures the distanceamongwords
and synsets in WordNet’s graph structure, such as by
counting the number of edges among synsets. If the words
are closer, the synsets are similar. It works in 2 stages : i)
Generating the Synsets. ii) Generating distractors for all the
Synsets.
2) Conceptnet: Conceptnet works well with Nouns and
Proper Nouns. Hence for the noun and proper noun type of
keywords Conceptnet isusedoverWordnet.Itisaknowledge
graph that connects words and phrases of natural language
with labelled edges. Its knowledge is collected from many
sources that include expert-created resources,
crowdsourcing, and games with a purpose.
Fig -2: Fill in the blanks
B. Numeric Fill in the Blanks
For generatingappropriatedistractorsforanumericanswer,
adding any integer between +5 and -5. Any random number
will be selected between this range and it will be added to
answer and corresponding three distractors would be
generated. Then shuffling these numbers using the
FisherYates Shuffle algorithm so that the answer won’t be
just a particular choice rather it would be a random choice.

Fig -3: Numeric Fill in the blanks
C. True or False questions
By using this algorithm true or false questions are generated
so only options for the question would be True or False so no
algorithm for generating the distractors is required.

Fig -4: True or False questions
D. Match the Following .
‘Sense2Vec’ is applied to generate the most sense options for
the filtered keywords as column 2 to match the following
pattern, ensuring to find the most identical word for the
filtered keywords. For Example : “ most sensewordforforest
is woods likewise for paws is back feet.” Framing the correct
option first, then the other distractors. To begin with, the
generated choices are a,b,c,d and used the loop to find the
appropriate option, i.e., the correct value for the filtered
keyword, and assigned the random choice, ensuring that
other keywords are assigned to other choices, and repeated
this process for the other filtered keywords. For example :
“woods gets the choice d as a random choice then the back
feet will be assigned to another choice but not d as it is
already assigned”. After generating the correct option, other
distractors are generated. Finally, the Fisher-Yates Shuffle
algorithm is used to shuffle these distractors so that the
output is not a definite choice but rather a random choice.

Fig - 5: Match the Following
5. RESULTS
The algorithms are implemented in Python using Flask as a
framework and tested on a number of different text files
which are taken from a variety of different websites which
includes blogs, Wikipedia articles, articles based on some
topics and so on. The experimental results show how
effective the system is, for extracting the MCQs from the text.
Following are some of the resultant MCQs that have been
obtained when providing the following text file.
Fig - 6: Input file
Output :
[
{
"question": "The most recent version forserver computers
is Windows Server _____ version 21H2",
"answer": "2022",
"id": 1,
"choice3": "2022",
"choice4": "2021",
"choice1": "2024",
"choice2": "2023"
},
{
"answer": false,

"question": "Microsoft Windows, commonly referred to as
Windows, is a group of multiple proprietary graphics
operating system families, all of which are notdevelopedand
sold by Microsoft",
"id": 2
},
{
"question": "A specialized version of Windowsalsorunson
the Xbox One and Xbox Series XS video ______ consoles",
"answer": "Game",
"id": 3,
"choice3": "Positivity",
"choice4": "Critical Mass",
"choice1": "Game",
"choice2": "Acting"
},
{
"question": "1) server → Just The Number 2) number →
Whole Game 3) game → Desktop Computer 4) computer →
Same Server ",
"answer": "1 → d 2 → a 3 → b 4 → c",
"id": 4,
"choice3": "1 → b 2 → c 3 → d 4 → a",
"choice4": "1 → d 2 → a 3 → b 4 → c",
"choice1": "1 → a 2 → d 3 → b 4 → c",
"choice2": "1 → d 2 → a 3 → c 4 → b"
},
{
"answer": true,
"question": "Still numbers for server use of Windows that
are comparable to competitors show one third market share
similar to that for end user use",
"id": 5
}
]
So, the system is evaluated manually where evaluatorscheck
the system on various text files to ensure the syntactic and
semantic correctness of the questions and also the quality of
distractors. It is observed that the number of MCQs that are
formed from any text are variable. This is directly
proportional to the summarization of the text. The exactness
of MQCs generator is found to be really good (nearly all
questions are of great level) since distractors are chosen by
applying various techniques. Hence, the framework
effectively creates the programmed different types of MCQs.
6. CONCLUSION
The system is tested by giving various inputs on different
domains in the form of text and the systemisworkingwellby
generating good quality MCQs and generates the syntactic
and semantic correct questionsalongwiththegoodqualityof
distractors. The problem of manually creating the MCQs is
solved, and the system is helpful for teachers for generating
the MCQ type questions.
Following are the future scopes of the system:
1) To build the same system for other languages such as
Hindi, Urdu, South Indian Languages, etc.
2) Various different types of multiple-choice questions like
answers containing images, a multiple choice question
containing question and multiple correct answers could be
added.
REFERENCES
[1] Annamaneni Narendra, Manish Agarwal and Rakshit
shah (2013) “AutomaticCloze-QuestionsGeneration”.In
Proceedings of Recent Advances in Natural Language
Processing, pp. 511–515.
[2] Ayako Hoshino, Hiroshi Nakagawa (2005). “A real-time
multiple-choice questiongenerationforlanguagetesting
a preliminary study”. In Proceedings of the 2nd
Workshop on Building Educational Applications Using
NLP, pp. 17–20.
[3] [3] Ibrahim Eldesoky Fattoh, 2014. “Automatic Multiple
Choice Question Generation System for Semantic
Attributes Using String Similarity Measures”. In
Computer Engineering and Intelligent Systems
www.iiste.org ISSN 2222-1719(Paper)ISSN 2222-2863
(Online) Vol.5, No.8, pp. 66- 73.
[4] [4] Santhanavijayan, A., Balasundaram, S.R., Hari
Narayanan, S., Vinod Kumar, S., and Vignesh Prasad, V.,
2017. “Automatic generation of multiple choice
questions for e-assessment”. In Int. J. Signal andImaging
Systems Engineering, Vol. 10, Nos. 1/2, pp. 54-62.
[5] [5] Ming Liu, Rafael Calvo, A., and Vasile Rus, 2012. “G-
Asks: An Intelligent Automatic Question Generation
System for Academic Writing Support”. In Dialogue and
Discourse 3(2), pp. 101–124.
[6] [6] Shivank Pandey, Rajeswari, K.C., 2013. “Automatic
Question Generation Using Software Agents for
Technical Institutions”. In International Journal of
Advanced Computer Research (ISSN (print):2249-7277

ISSN (online): 2277-7970) Volume-3 Number-4 Issue-
13, pp. 307-311
[7] [7] Manish Agarwal, Rakshit Shah and Prashanth
Mannem, 2011. “Automatic Question Generation using
Discourse Cues”. In Proceedings of the Sixth Workshop
on Innovative Use of NLP for Building Educational
Applications, pp. 1–9.
[8] [8] Arjun Singh Bhatia, Manas Kirti, and Sujan Kumar
Saha,2013. “Automatic Generation of Multiple Choice
Questions Using Wikipedia”. In P. Maji et al. (Eds.):
PReMI 2013, LNCS 8251, pp. 733–738.

Vikalp - Automatic multiple choice questions generator

Recommended

More Related Content

Similar to Vikalp - Automatic multiple choice questions generator (20)

More from IRJET Journal (20)

Recently uploaded (20)

Vikalp - Automatic multiple choice questions generator