An Early Warning Detection System of Terrorism in Indonesia from Twitter Contents using Naïve Bayes Algorithm

Abstract

Aware on the benefits of social media as the networking platform, the extremist organization is utilized social media to spread the ideology, recruit new member and guided a suicide bomber alike. There are opportunities to analyze the content of document texts in social media including the terrorism detection and intention by extracting the content evident in their post, comment etc. The objective of this research is to analyze content posted in Twitter and to review whether post and conversation on Twitter will be highly related to terrorism intention or another way around. This study deployed Naïve Bayes classification technique which identified Twitter contents in Indonesian national language. The method has been processed text pre-processing, and dataset divided with hold out technique. Result of F-measure value indicates that 76% and 77% of texts are associated with the accuracy level of terrorism based on macro-averaging and micro-averaging indicators. The finding is contributed to the scanty literature on the early warning detection method in Indonesian language and assist the government to target the extremists' organizations.