Pro-ISIS fanboys network analysis and attack detection through Twitter data

Terrorism becomes more severe these days, especially the attacks sponsored by Islamic State of Iraq and Syria (ISIS) or Daesh. They are experts of using social network for propaganda and recruitment, thus predicting their activities through big social network data will help to predict and avoid more serious attacks. In this paper, we analyze over 17k Twitter records of pro-ISIS fanboys over a year. Based on those tweets, we want to dig out: 1. Who is the most important and active member in the social network? 2. What are the hashtags they frequently used for propaganda? 3. According to the twitting peak time and information on-hand, will it be able to predict the next attack? Our results reveal the leader in this propaganda network, and achieve a satisfactory attack prediction via time-series neural network through very limited attack history.