Using Machine Learning for Arabic Sentiment Analysis in Higher Education: Investigating the Impact of Utilizing the ChatGPT and Bard Google
Gaining an understanding of the application's quality and meeting, the user's needs are crucial in the development of applications. To enhance the quality of applications, it is important to comprehend the requirements of the users. One effective approach for achieving this is th...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | Arabic |
| Published: |
Thamar University
2025-03-01
|
| Series: | مجلة العلوم الهندسية والتقنية |
| Subjects: | |
| Online Access: | https://journal.tu.edu.ye/index.php/Joeats/article/view/2473 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Gaining an understanding of the application's quality and meeting, the user's needs are crucial in the development of applications. To enhance the quality of applications, it is important to comprehend the requirements of the users. One effective approach for achieving this is through the utilization of application review-based sentiment analysis (SA). In this study, the objective was to assess students’ opinions regarding mobile applications of universities in order to update and maintain them accordingly. Mobile applications of universities have become an integral part of students’ lives, thus making it imperative to analyze user comments on these apps for SA purposes, where student input is crucial for assessing the effectiveness of educational institutions. This paper presents a machine learning (ML) based approach to sentiment analysis on students’ evaluation of higher education institutions. The study analyzes a corpus containing approximately 275 student reviews written in Arabic, It also evaluates the performance of three ML techniques, including K-Nearest Neighbors (K-NN), Decision Tree (DT), and Random Forest (RF) using an accuracy, precision, recall, and f-score measures. In addition, the study compares one method of labeling the data for ASA, including manual labeling by humans, labeling by Bard Google and labeling by ChatGPT. Experimental results show that the K-NN technique performed the best, achieving an accuracy of 74.91% by ChatGPT models for Arabic sentiment analysis (ASA). Moreover, utilizing proposed active labeling method with Bard Google achieved higher accuracy compared to other labeling methods. The study proposed study suggests that the K-NN technique with ChatGPT models and proposed active labeling method are effective approaches for ASA by ChatGPT. It is indicated by the empirical results that promising results are yielded on the evaluation of students' opinions of higher educational institutions by an ML based approach.
|
|---|---|
| ISSN: | 2958-809X 2958-8103 |