Veil and Hijab: Twitter Sentiment Analysis Perspective

Abstract

Controversies about veil and hijab are often occur in society. Especially in today’s digital era, public opinion expressed through social media can greatly influence the others opinions, regardless of whether it is positive or negative. Therefore, this research was aiming to conduct an approach through analysis sentiment of public opinion about the veil and hijab to know how much accurate the sentiment analysis predict the positive, negative, or other sentiments with using Twitter data as the research object. The algorithm used in this study is Support Vector Machine (SVM) because of its fairly good classification model though it trained using small set of data. The SVM on this research was combined with Radial Base Function (RBF) kernel because of its numerical difficulties that are fewer than linear and polynomial kernel and also because this research doesn’t have a large feature. The amount of data used is 3556 tweets data. Tweets data, which is numbered 1056, is classified manually for the learning process. The remaining 2500 data will be classified automatically with the classifier model that has been created. A total of 1056 tweets data that have been classified manually is separated into training and testing data with a ratio of 8: 2. The result of the sentiment analysis process using Support Vector Machine algorithm RBF kernel with C=1 and γ=1 has an accuracy score of 73.6% with precision to negative opinions are 62%, positive opinions are 83%, neutral opinions reach 53% and irrelevant opinions that talk about hijab and veil reach 98%. It shows that sentiment analysis can be used for predicting the negative, positive or other sentiments of a sentence based on a certain topic, in this case veil and hijab.