A Study of Radicalism Contents Detection in Twitter: Insights From Support Vector Machine Technique

Social media has been widely used to target, coordinate, disseminate and embed radicalism doctrine to the society. Radicalism gives destructive impacts to an attacked country and the global society. It is not only makes a serious threat to a nation unity, yet it also impacts the cultural, socio-political, and economic among the others. This study is aim to study radicalism intention using content detection. The model was developed using social media content like posting, comment and conversation to indicate the level of radicalism. Data was collected from a Twitter. It has chosen as a social media platform because Twitter can engage the community with powerful and impactful microblogging function. The data was analyzed using machine learning. To be specifically, a Support Vector Machine (SVM) in text mining employed for classification of Twitter's content in national language of republic Indonesia. There were two focus keywords used in this study. The first one is ISIS and follow by Syria. The result shows that 83.3% accuracy of test set tuples, 90% part of the contents indicates positive link to radicalism with no-radicalism class as a precision value. Besides recall value was calculated for accuracy, 95% and 82% part of actual text positively link as radicalism class with a proxy of no-radicalism class. This model is hoped to detect and overcome the terrorism issue in Indonesia.