To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Sentiment Analysis of Twitter Data on Smart City Services using BERT Model |
|
Author |
Abrar Albeladi and Turki Alghamdi |
| Citation |
Vol. 25 No. 2 pp. 64-80 |
|
Abstract |
Sentiment analysis of online platforms has emerged as a significant research area in various domains, including smart cities. Governments in many countries aim to improve the quality of service provisions to ensure sustainability and create smart city systems. The use of technologies to support this goal makes cities a great source of data. Social media platforms, particularly Twitter, offer a rich source of data, as users share their opinions, emotions, and evaluations about events and services. Governments can utilize these citizens’ opinions to enhance services provisions. However, the abundance of noisy data on social media platforms poses a challenge in identifying relevant information, hindering timely responses to citizen feedback. This study focuses on analyzing Arabic tweets related to smart cities from two Saudi cities to assist decision-makers in future planning. A three-class dataset with each class containing 1109, 771, and 8804 samples was used for experimentation. Two sets of experiments were conducted: one without Synthetic Minority Oversampling Technique (SMOTE) and the other with SMOTE. The experimental results without SMOTE revealed that linear support vector machine (SVM) and multinomial Naive Bayes (NB) achieved better performance. However, more promising results were obtained with SMOTE analysis, where Logistic Regression, Random Forest, Linear SVC, Gradient Boosting, Decision Tree, Voting, and Stacking algorithms demonstrated superior performance. |
|
Keywords |
Sentiment analysis, Random forest algorithm, Smart cities, Arabic tweet datasets, SMOTE |
|
URL |
