|
Document Type |
: |
Thesis |
|
Document Title |
: |
AUTOMATIC CLASSIFICATION OF TOPICS POSTED AT SAUDI UNIVERSITIES IN TWITTER التصنيف التلقائي للمواضيع المتداوله في حسابات الجامعات السعودية في تويتر |
|
Subject |
: |
Faculty of Computing and Information Technology |
|
Document Language |
: |
Arabic |
|
Abstract |
: |
Social media are the key venue of social interaction today. One such medium is Twitter whose characteristics and use has become the focus of various fields such healthcare, communications, and education. As education is the main field for the development of societies, educational institutions have paid attention to measuring their interaction with their students on Twitter to improve the performance of educational services by classifying tweets. Classifying tweets via machine learning algorithms is a viable means of evaluating the effectiveness of social media campaigns and communications. The present study is aimed at examining the field of social network mining through classifying the tweets in Arabic of various sectors of King Abdul-Aziz University, an educational institution in Saudi Arabia, through building a framework to classify the tweets into specific categories such as admission, exams, and libraries. This framework examines the power of stylometric features using three different classifiers of Arabic microblog. The stylometric features applied to the text in this framework are character, lexical, and syntactic. These features were extracted using the MADAMIRA tool and were used to build a features-based classification model. In this model, the features are combined with a different size set of data using three classification techniques, which are support vector machine (SVM), k-nearest neighbour, and random forests. The K-folds technique was used to evaluate the models, and the measures of precision, accuracy, recall, and f-measure were made regarding every machine-learning model. The study compared the three models to ascertain which models would offer the best performance in terms of accuracy. The findings show that SVM is the best performing classifier. While the random forest also proved to be a strong classifier, it did not perform as well as the SVM. |
|
Supervisor |
: |
Dr Naif Radi Aljohani |
|
Thesis Type |
: |
Master Thesis |
|
Publishing Year |
: |
1442 AH
2020 AD |
|
Added Date |
: |
Saturday, February 6, 2021 |
|
Researchers
| ولاء صالح الحبشي | Alhebshi, Walaa Saleh | Researcher | Master | |
|