Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

St.

Mary’s University
School of Graduate Studies
Faculty of Informatics
Department of Computer Science
Course Natural Language pressing

Reviewed on Automatic Generation of Amharic Math


Word Problem and Equation

Reviewed by: ID NO.

1. Brtukan Hussein SGS /0526/2013A

2. Michael Asrat SGS /0386/2013A

3. Nuguse Negese SGS/0464/2013A

4. Mikias Melaku SGS /0507/2013A

5. Yared Wolderufael SGS /0396/2013A

SUBMITTED TO: DR. Michael Melese (PHD)


SUBMISSION DATE: DECEMBER 23, 2021
CHAPTER ONE

Abstract
Math word problems are an important part of math learning. Students must be exposed to new
and more math word problems in order to successfully develop problem-solving skills and improve
their reading comprehension. Automatic generation of math word problems has been proven to
be successful in providing new and varied exercises to meet the needs of each individual student.
As a result, numerous research projects on the automatic generation of math word problems and
equations for the English language and a few other languages have been proposed. Amharic is
a language spoken by over 100 million Ethiopians and is used as a medium of instruction in
primary schools in Ethiopia. Taking advantage of this opportunity to address such a large
population, I proposed in this paper an automatic generation of AMW problem and equation using
a template-based shallow NLP approach. The proposed method accepts sample AMW problems
to create a template that is then used to generate AMW problems. Three experiments were carried
out to validate the proposed approach. The experimental results show that the proposed approach
has an overall efficiency of 93.84 percent when formulating AMW problems and equation
templates. Experiment has also shown a 90.24 percent relatedness to human generated AMW
problems and has also proven to generate solvable problems as it has shown a 90 percent
accuracy of giving correct answer when compared to human solver answer. Finally, while the
proposed strategy has shown promising results when using a template-based shallow NLP
technique, it has limitations when it comes to generating AMW problems with deeper semantic
meanings. As a result, this research work paves the way for future research on the use of deeper
NLP techniques to investigate more efficient strategies.
Introduction
According to one study, using NLP for automatic generation of math word problems with their
equation can provide students with the opportunity to practice on more problem-solving exercises
tailored to their specific needs, thereby improving their problem-solving ability [2]. As a result,
quite a few research works on techniques to generate math word problems and equations
automatically for the English speaking community as well as for a few other languages have been
proposed. In this paper, I present a strategy for automatically generating math word problems and
equations in Amharic.

Natural language is a fundamental aspect of human behavior. It is an essential component for


communicating about the world and how it affects our daily lives. Because the majority of human
knowledge is recorded using natural languages, only computers with the ability to understand
natural language can efficiently access the information contained in the natural language. Natural
language processing (NLP) refers to computers' ability to generate and interpret natural language.
NLP is also a significant area of study in computer science. When NLP capabilities are fully
realized, computers will be able to understand and process natural language, translate languages
accurately and in real-time, or extract and summarize information from a variety of data sources,
depending on the users' requests. Afro-Asian languages are subdivided into the Semitic branch.
More than a hundred million Ethiopians speak Amharic, a Semitic language. It is the second most
spoken Semitic language after Arabic. It is the official working language of the Federal Democratic
Republic of Ethiopia, giving it official status throughout the country. It is also the official or working
language of a number of federal states/regions, including Amhara and the multi-ethnic Southern
Nations, Nationalities, and Peoples region. Amharic is the language of millions of emigrants
worldwide, and it is also spoken in Eritrea [3]. It is one of the most common languages taught in
Ethiopian elementary schools. It is written in a writing system called Fidel or abugida, which was
adapted from the now-extinct Ge'ez language. Because Amharic is a morphologically rich
language, it poses a challenge to the field of natural language processing in general [4].
Despite the fact that Amharic is a less resourced language in the field of linguistics and natural
language processing (NLP), several baseline NLP research works on the Amharic language have
been completed so far. For instance, POS tagging and Morphological analyzer [5, 6], Spell
checking, and named entity recognition. This NLP research work paves the way for future
research on the Amharic language. As a result, this paper builds on these studies to present a
proposed automatic generation of AMW problems and equations using a template-based shallow
NLP approach as a step toward enabling comprehension and learning problem solving in
mathematics for elementary level students.
The template created in the first phase is used to generate AMW problems and equations. The
generator subcomponent reads the template from the database and begins the generation of
AMW problems and equations. This process of creating an AMW problem and equation is
followed by the solver subcomponent, which solves the equation. Finally, the engine saves the
generated word problem along with the solution to a database. The following sections explain this
phase in detail as a two-step process.
Challenges
Math word problems are a type of mathematics exercise in which the main background
information for the problem is presented in textual words rather than mathematical equations.
Learners gain basic linguistic knowledge as well as skills in basic arithmetic operations such as
addition, subtraction, multiplication, and division through math word problems. Students can solve
normal mathematical operations very well, but they struggle when these operations are hidden
behind word problems in text form.
Despite the fact that the teacher's assistance cannot be ignored, the teacher cannot provide all
types of questions and does not provide a sufficient volume of exercises that take into account
individual differences among students. Furthermore, textbook math word problems lack variety,
and students may grow tired of solving the same types of problems over and over.

Word problems are also known as story problems because they frequently include a narrative of
a real-world incident, and the amount of technical terminology used varies.

Statement of the problem and gap


To master math word problem solving, students must be exposed to a large number of new
problems, which most textbooks and teachers fail to provide.

Objective of the study


As a result, in this paper, a strategy for automatic generation of Amharic Math Word (AMW)
problem and equation is proposed, as a first attempt to introduce the use of template-based
shallow NLP approach to generate math word problems for the Amharic language as a step
towards enabling comprehension and learning problem-solving in mathematics for primary school
students. To create a template, the proposed novel technique accepts a sample AMW problem
as user input.

Methodology
It is used as a pattern to generate AMW problems with semantically equivalent equations. A
prototype was created and used as a testing platform to validate the feasibility of the proposed
approach.
The researcher proposed an automatic generation of AMW problems and equations using a
template-based shallow NLP approach in this paper. The proposed method accepts sample AMW
problems to create a template that is then used to generate AMW problems. Three experiments
were carried out to validate the proposed approach. The experimental results show that the
proposed approach has an overall efficiency of 93.84 percent when formulating AMW problems
and equation templates. Experiment has also shown a 90.24 percent relatedness to human
generated AMW problems and has also proven to generate solvable problems as it has shown a
90 percent accuracy of giving correct answer when compared to human solver answer.
Though the use of appropriate teaching methodology can assist them in reducing this challenge,
the power of technology must be harnessed to generate a fresh and massive amount of math
word problems for students' unrestricted practice in order to address the aforementioned
challenge from the ground up.

Weakness and Strong


Weakness
Even though the paper discussed the Automatic Generation of Amharic Math Word Problems and
Equations, it attempted to solve only using Amharic and no other Ethiopian languages.
Furthermore, they made no recommendations for the next researcher in the field. Some
annotation methods were not compared or recommended in the paper.

Strength
As far as I can tell, the strongest and most positive aspect of the paper is that it serves as a
guide for those who want to do more research on the subject.

The author experimented with intra-annotator agreement, attempting to detect errors by


identifying "outliers" with non-uniform behavior.

Limitation
It is limited in its ability to generate AMW problems with deeper semantic meanings. As a result, this
research work paves the way for future research on the use of deeper NLP techniques to investigate
more efficient strategies.

Future Scope
Furthermore, they are working on the design and development of our own novel routing protocol
that will be the best in terms of routing performance, end-to-end delay, packet overhead, and
throughput based on ACO- Swarm Intelligence Technique.
Conclusion
Furthermore, they are working on the design and development of our own novel routing protocol
that will be the best in terms of routing performance, end-to-end delay, packet overhead, and
throughput based on ACO- Swarm Intelligence Technique.
Due to this challenge, the subject math word problem solving necessitates the teacher's
increased attention to student activities in order for students to receive an abundance of new
math word problems to solve as well as proper problem-solving assistance.
A strategy for automatic generation of Amharic Math Word (AMW) problem and equation is
proposed in this paper, which is a first attempt to introduce the use of template-based shallow
NLP approach to generate math word problem for Amharic language as a step towards enabling
comprehension and learning problem solving in mathematics for primary school students. To
create a template, the proposed novel technique accepts a sample AMW problem as user input.
A template contains an AMW problem with placeholders, a problem type, and an equation
template.
It is used as a pattern to generate AMW problems with semantically equivalent equations. A
prototype was created and used as a testing platform to validate the feasibility of the proposed
approach.

Reference
[1] Word Problem.
https://en.wikipedia.org/wiki/Word_problem_(mathematics_education)
[2] Nandhini, K. and Balasundaram, S.R. (2011) Techniques for Generating Math Word
Questions for Learning Disabilities. IJCA Proceedings on International Conference and
Workshop on Emerging Trends in Technology (ICWET), No.4, 45-50.
[3] Amharic Language.
http://www.lonweb.org/link-amharic.htm
[4] Dehadari, J., Tounsi L. and Genabith, J. (2011) Morphological Features for Parsing
Morphologically-Rich Languages: A Case of Arabic. Proceedings of the 2nd Workshop
on Statistical Parsing of Morphologically Rich Language (SPMRL 2011), Dublin, Ireland,
October 2011, 12-21,
[5] Gasser, M. (2011) HornMorpho: A System for Morphological Processing of Amharic,
Oromo, and Tigrinya. Conference on Human Language Technology for Development,
Alexandria, 2-5 May 2011, 94-99.
[6] Alemu, B. (2013) A Named Entity Recognition for Amharic. MA Thesis, Addis Ababa
University, Addis Ababa.

You might also like