A Sentiment Analysis Using Fuzzy Support Vector Machine Algorithm

Abstract

The Ministry of Communication and Information and the Ministry of BUMN of The Republic of Indonesia designed a mobile app “Peduli Lindungi” to be used to help the public and related government agencies in carrying out screening and tracing people's movement to stop the spread of Corona Virus Disease (Covid-19).The existence of a mobile app, “Peduli Lindungi” triggers abundant different sentiments from the Indonesian community, either positive or negative sentiments. Based on the positive sentiment, the government of the Republic of Indonesia may have some feedback about the aspects of the app that should be maintained. In contrast, negative sentiments can be used as initial points of the potential improvement of the mobile app. This study applies a Fuzzy Support Vector Machine (FSVM) model to classify the user's reviews on Peduli Lindungi Application. FSVM can classify customers’ reviews into two or more classes and relatively results in higher accuracy than other classification approaches. The results of this study indicate that the classification of reviews with FSVM produces quite good accuracy with a value of 77%. A total correct prediction is 2192 reviews out of 2813 reviews.