Twitter Sentiment Analysis on Activities of Saudi General Entertainment Authority

Abstract

Sentiment analysis can be defined as a natural language process to determine the individual’s sentiment or opinion towards something. It helps institutions, companies and governments to gain a deeper understanding and supports decision-making. This paper aims to analyse individuals’ opinions in Twitter on the activities of the Saudi General Entertainment Authority (GEA) using machine and deep learning techniques. To achieve this aim, 3,817 tweets were collected using RapidMiner. To classify tweets into supporters and opposers, three machine learning algorithms were used namely, Multi-Layer Percptron (MLP), Support Vector Machine (SVM), Random Forest (RF), and one deep learning algorithm, which is Recurrent Neural Network (RNN). Two test options were applied to evaluate the classification model, percentage split and K-fold validation tests. The results show that the people are happy and agree with the GEAs’ activities. As for the gender, the support rate of females was higher than males. In addition, RF algorithm outperforms other algorithms in terms of the classification accuracy and the error rate.