Scribd
Upload
18 views3 pages

A Review On Various Techniques For Automatic Question Generation

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) Web Site: www.ijettcs.org Email: [email protected] Volume 3, Issue 6, November-December 2014 ISSN 2278-6856

Uploaded by

International Journal of Application or Innovation in Engineering & Management
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
18 views3 pages

A Review On Various Techniques For Automatic Question Generation

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) Web Site: www.ijettcs.org Email: [email protected] Volume 3, Issue 6, November-December 2014 ISSN 2278-6856

Uploaded by

International Journal of Application or Innovation in Engineering & Management
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]


Volume 3, Issue 6, November-December 2014

ISSN 2278-6856

A Review on various Techniques for Automatic


Question Generation
Ramandeep Kaur1, Shilpy Bansal2
1

M.Tech Final Year Student, Department of Computer Science &Engg


Bhai Maha Singh College of Engineering & Technology, Sri Muktsar Sahib, Punjab, India
2

Assistant Professor,Department of Computer Science &Engg


Bhai Maha Singh College of Engineering & Technology, Sri Muktsar Sahib, Punjab, India

ABSTRACT
This paper presents a review of generating questions
automatically. It has its potential usage is in education field.
Not much work has been done in this field of question
generation for Punjabi language in recent years. While setting
exercises questions, teachers use test markers to construct
questions bank for examination. By automatically generating
the Questions, it decreases the time spent on constructing
papers.For Punjabi, Semantic Role Labeller, NER, POS
Tagger, annotated corpora tools are not yet available in
required manner. Various Names Entity Recognition and
dictionary lookup approaches can be used to develop desired
output.

Keywords:- Question Generation System, NER, Semantic


Role Labeller, Punjabi Language

1.INTRODUCTION
Question generation is a strategy that assists learners with
their comprehension of text. Students learn to formulate
and respond to questions about situations facts and ideas
while engaged in understanding a text. As computers
plays an important role in each field, like Education field,
And language learning field, there are so many tools have
been developed already and play an important role in
learning fields. They have the ability to ask questions for
like: What is your name? etc. We explained it with the
help of following example:
Input
Sentence:

Volume 3, Issue 6, November-December 2014

Automatic Question generation becomes quite beneficial


because it generate the questions automatically. A person
can ask a question on the web and receives answer in a
few seconds. Many years ago, it was very difficult task. At
that time it will take hours or weeks to receive the answers
the same question. The new technology will radically
transform how we think, behave and react. The purpose of
asking questions is not limited, in a classroom; a teacher
asks the question to his students because to actively
engage the students in lesson, to create interest in the
topic and generating such type of questionsthat helps to
develop the students critical thinking abilities. Automatic
question generation systems can produce test questions up
to 4 times faster than a human expert, without
compromising quality [1]. Various applications of
Question generation are in Intelligent Tutoring system,
closed domain question Answering and natural language
summarization. To learners by suggesting good questions
while reading documents. In medical field Suggesting
questions foe parents & caretaker. Togenerate questions
regarding security context and legal context [2]. The task
of question generation can be viewed as a three step
process- Content Selection, Selection of question type and
Question
construction.Content
Selectioncomponent
determines which information should be conveyed in the
output. Question generations very first task is to decide
which pieces of information to convey in the output.
Content selection is quite domain dependent. Selection of
question type is depends on two premises. First, it is a
suitable question if it revolved around an important
concept in the paragraph or given sentence. Secondly, a
question was selected if the same or similar question was
also generated by one or other systems like Human Expert.
For example, the question What is Computer? was
selected as a question because it revolves a particular
concept and the same type of question was generated by
Human expert. Question construction is the process in
which questions are asked to obtain useful information
about a particular topic.Question generation, is the process
that encapsulates text processing, scoring and question
Overgeneration that encompasses this systems task of
automatically generating questions from input.
Page 104

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)


Web Site: www.ijettcs.org Email: [email protected]
Volume 3, Issue 6, November-December 2014

Tool for Question Generation


Names Entity Recognition:The term named entity
recognition, widely used in Natural Language Processing.
The computational researches are performed to
automatically identifying names entities in texts from a no
of methods and representations. Performance of any NER
system relies on heuristics and handcrafted rules. There is
no proper corpus is available for Punjabi language hence
there is a need to develop a NER system for Punjabi
language. Named entities are basically atomic elements in
text that belongs to predefined categories such as the
names of persons, organizations, locations, expressions of
times, quantities, monetary values, percentages, etc."
Named entity recognition (NER) is the task of identifying
such named entities. Named entity recognition, although a
seemingly simple task, faces a number of challenges.
Entities may firstly be difficult to find, and once found,
difficult to classify. Locations and person names can be
the same, and follow similar formatting. Named Entity
Recognition system is very useful in various applications
like automatic question generation system, text
summarization system etc. Hence to make accurate
question generation system an effective NER System must
be developed.
Corpus for Punjabi: When we recognize the crucial value
of an error free corpus with a vast number of examples in
solving natural language processing tasks, the task of
building a corpus with these properties gains importance.
The most prominent method of building corpora today is
to divide the work among experts and wait for them to
finish work. A proper corpus for Punjabi language is
required to develop an effective question generation
system for Punjabi Language.
EXISTING SYSTEM
The history of NLP generally starts in 1950s, although
work can be found from earlier periods. Research in
natural language processing has been going on for several
decades back. Machine translation was the first computer
based application related to natural language. Garg and
Goyal[3] introduce a system that is based on Named
Entity Recognition, Various Punjabi language dependent
rules to generate outputs. The system uses various
manually designed named entity lists to transform the
declarative sentences into a set of possible questions. This
system generate the questions starts with the words like

Volume 3, Issue 6, November-December 2014

ISSN 2278-6856

Kitthe (Where) Konn (Who)etc. Heilman [4]


introduces a system which uses the combination of manual
rules and statistical question ranker trained on a dataset of
system output. Heilman introduces a system that is based
on generating factual questions with an Overgenerationand-Ranking strategy based on NLP techniques. He also
uses the Name Entity Recognizer, WH-movement, Subject
auxiliary
inversion
for
generating
factual
questions.Mannem et al., [5] develops a system in which
questions are generated from paragraphs rather than
sentences. The system uses SRL (Semantic Role Labels)
that is provided by ASSERT. This provides multiple
predicate arguments structure for sentences. He uses three
stage processes for question generation, Content selection,
Question formation and Ranking.Aggarwal et al., [6] uses
Discourse Cues in the system that the Discourse relations
help to analysing the sense of the connectives that help in
generating questions of Which, What, When etc. Xu et al.,
[7] introduce a system Intelligent Tutoring system that
interacts via speech. In that process the learners reads
and several statements, listen to questions that is
generated by the system. The system then checks the
correctness of each answer. Lemos et al., [8] proposed an
alternate method to generate questions from template.
When user makes a request to the template, a new
instance of the question is generated. There are also many
other researches were done for question generation
system. Recently Kunichiks et al., [9] develop a
multimedia language learning system for teaching English
to Japnese students.
Conclusion and Future Work
Our approach is to generate questions from text. For this
purpose we discuss a no. of previous techniques and tools
that are used for question generation process. Some
techniques are based on Text2 Test based questions
approach while others are on questions from paragraphs.
Many techniques and tools are available for English
language but for Punjabi language very few tools are
available. To generate questions in Punjabi there are lots
of language dependent rules and also a problem of nonavailability of large gazetteer. The existing systems are
able to generate only shallow questions with the question
words (what), (When), (How many/ how much).
Future work includes a no. of new rules and techniques to
improve the existing ones. In Punjabi, many first names in
Punjabi are also common nouns. This will also degrade
the system performance. In case of multiple sentences in
the input, system is not able to identify any relation
between them.

References
[1] Y. Skalban,R. Mitkov, L. Specia, Automatic
Question
Generation
in
multimedia-based
learning,In Proceedings of the 24th International
Conference on Computational Linguistic, pp. 1151
1160, 2012.
[2] V.Rus, A. C Graesser, The Question Generation
Shared Task and Evaluation Challenge,Dialogue and
Discourse 3 (2), pp. 177-204, 2012.
Page 105

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)


Web Site: www.ijettcs.org Email: [email protected]
Volume 3, Issue 6, November-December 2014

ISSN 2278-6856

[3] S. Garg ,V. Goyal, Automatic Question Generation


from Given Punjabi Text, International Journal of
Emerging Trends & Technologies in Computer
Science 3(4), pp. 81-96, 2013.
[4] M. Heilman, Automatic Factual Question Generation
from Text, PhD Thesis, Carnegie Mellon University,
2011.
[5] P.Mannem, R. Prasad, A. Joshi Question Generation
from Paragraphs at UPenn: QGSTEC System
Description,In Proceedings of the Third Workshop on
Question Generation,2010.
[6] M. Aggarwal, R. Shah,P. Mannem, Automatic
Question Generation from Discourse Cues, In
Proceedings of the Sixth Workshop on Innovative use
of NLP for Building Educational Applications, pp. 19, 2011.
[7] Y. Xu, A. Goldie, S. Sneff, Automatic Question
Generation and Answer Judging: A Q&A Game for
Language Learning,MIT Computer Science and
Artificial Intelligence Laboratory, United States,
2009.
[8] M. Lemos, R. Muralidharan, V.V. Kamat,
Automatic Generation of Questions for On-line
Evaluation,In Proceedings of National Seminar on
E-learning & E-learning Technologies, Hyderabad,
2001.
[9] H. Kunichika, T. Katayama,A. Takeuchi, Automated
Question Generation Methods for Intelligent English
Learning Systems and its Evaluation, In Proceedings
of the International Conference on Computers in
Education, pp.1117-1124, 2001.

Author[s] brief Introduction


Ramandeep Kaurreceived B. Tech degree in Information
Technology& Technology from Swami Vivekanand
College of Engineering & Technology, Banur
(Chandigarh) in 2012. Currently she is pursuing her M.
Tech from BMSCET, Sri Muktsar Sahib, Punjab. Her
research interest includes Natural Language Processing.
Er.Shilpy Bansal is currently working as Assistant
Professor in the Department of Computer Science &
Engineering, BMSCET, Sri Muktsar Sahib, Punjab.

Volume 3, Issue 6, November-December 2014

Page 106

You might also like