Scribd
Upload
79 views3 pages

Assignment 1 NLP

1. The document introduces the Superior Arabic Text Categorization Deep Model (SATCDM), which uses a convolutional neural network and word embeddings to classify Arabic text documents into predetermined categories. 2. SATCDM aims to improve the accuracy of classifying Arabic news articles, as traditional machine learning approaches have difficulties with the complex structure and morphology of the Arabic language. 3. The model is evaluated on 15 freely available datasets of Arabic news text in Modern Standard Arabic, achieving classification accuracies ranging from 97.58% to 99.90%, outperforming other models.

Uploaded by

Roshini Rames
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
79 views3 pages

Assignment 1 NLP

1. The document introduces the Superior Arabic Text Categorization Deep Model (SATCDM), which uses a convolutional neural network and word embeddings to classify Arabic text documents into predetermined categories. 2. SATCDM aims to improve the accuracy of classifying Arabic news articles, as traditional machine learning approaches have difficulties with the complex structure and morphology of the Arabic language. 3. The model is evaluated on 15 freely available datasets of Arabic news text in Modern Standard Arabic, achieving classification accuracies ranging from 97.58% to 99.90%, outperforming other models.

Uploaded by

Roshini Rames
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Fakulti Teknologi Maklumat dan Komunikasi

Universiti Teknikal Malaysia Melaka


BITI 3413 Natural Language Processing
Assignment 01 (Individual)

Lecturer Name: Ts. Dr. HALIZAH BINTI BASIRON


Group Members: -
No. NAMA NO. MATRIK
1. ROSHINI A/P RAMES B032110385
2. HERISH CHAUDRAY A/L RAVANTHERAN B032110409
INTRODUCTION

The research article “Superior Arabic Text Categorization Deep Model (SATCDM)” by M.
Alhawarat, Member IEEE and Ahmad O. Aseeri presents a deep learning methodology along
with Natural Language Processing (NLP) for classifying Arabic text documents. In the
domain of NLP, document classification is a crucial task that involves categorizing
documents into predetermined classes based on their content especially when it comes to the
classification of Arabic text documents is regarded as a significant area of study where the
number of Arabic documents is growing drastically in a daily basis due to increase in new
online pages and social media posts. Thus, it is crucial for users and future researchers to
categorise such documents into distinct types. Deep learning techniques and traditional
machine learning approaches are used to extract meaningful patterns from complex data for
an accurate classification of Arabic text documents.
Classifying Arabic documents shows a unique challenge due to the inherent complexities of
the language. This is because the Arabic Syntax is complex with different word orders and
grammatical structures than English which makes it difficult for traditional Machine Learning
(ML) algorithms to learn the patterns that distinguish different classes of Arabic text.
Secondly, the rich dialects and morphology of the Arabic language
Although deep learning techniques show great potential for obtaining even greater levels of
accuracy, classic machine learning (ML) techniques have been applied efficiently for this task
of Arabic text categorization.The study introduces the Superior Arabic Text Categorization
Deep Model (SATCDM), a novel deep learning methodology leveraging CNN and word
embedding. While deep learning has excelled in Computer Vision and Speech Recognition,
its application to Arabic Natural Language Processing (ANLP) is an ongoing area of
improvement. The SATCDM model, employing an efficient multi-kernel CNN architecture
and skip-gram word embedding with sub-word information, aims to enhance the accuracy of
classifying Arabic news text documents. The research employs 15 free datasets in Modern
Standard Arabic (MSA) format for evaluation, comparing results with traditional ML
techniques as baseline models. The outcomes are anticipated to significantly contribute to
ANLP by improving the precision of classifying Arabic text documents, thereby enhancing
search engine accuracy and other applications. The study is distinctive for being the first to
utilize word embedding and CNN for classifying Arabic news text in MSA format across
various freely available datasets. The article concludes with a comprehensive structure
covering a literature review, CNN introduction, data description, SATCDM model details,
methodology, experimental setup, results, and a concluding section.
In this study, the proposed model incorporates Convolutional Neural Network (CNN) and n-
gram word embedding to enhance classification accuracy. Despite the remarkable
advancements of deep learning in other fields, its application to the Arabic language,
especially in natural language processing, has been gradually improving. The SATCDM
model achieves high accuracy, ranging from 97.58% to 99.90%, surpassing similar studies in
Arabic document classification. The research employs 15 freely available datasets
representing Arabic news text documents in Modern Standard Arabic format. The comparison
includes baseline models using traditional Machine Learning (ML) techniques. The outcomes
are expected to significantly contribute to the accurate classification of Arabic text
documents, benefiting search engine retrieval and various applications in the field of Arabic
Natural Language Processing (ANLP) and Machine Learning (ML).

You might also like