Sentiment analysis of public towards infrastructure development in Indonesia on Twitter media using boosting support vector machine method

Abstract

Infrastructure is one of the important elements which supports the economic development of a country. Increasing the quality of infrastructure will also increase the level of the economy of a country. Infrastructure development also provides easy access for goods distribution or services throughout Indonesia and facilitates connectivity between communities. Besides, parallel infrastructure development is also needed to avoid economic inequality between regions in Indonesia. The development of social media in Indonesia has made people easier to convey their responses regarding the performance of government which one of them is related to infrastructure development. One of the types of social media that is currently popular among Indonesian people is Twitter. Considering that the popularity of Twitter in Indonesia, sentiment analysis as an application from text mining was conducted. We use support vector machine (SVM) as an analytical method to make a prediction model for classification of data opinion on Twitter. Some opinions which derived from it were classified into two groups, negative and positive opinions. The data was unbalanced data between negative class and positive class. Therefore, to handle this, a boosting algorithm on the SVM method was added. As a result, sentiment analysis of infrastructure development in Indonesia using the boosting support vector machine (BSVM) method gives an accuracy of 95.43%, a specificity of 81.67% and sensitivity of 97.54%. The classification results were made in the form of web framework application with visualization in the form of wordcloud and streamgraph graphics that were displayed interactively with a web-based application R Shiny.