Cybercrime profiling: Text mining techniques to detect and predict criminal activities in microblog posts

The exponential development in online social media allows users around the globe the possibility to share and communicate information and ideas freely in different formats of data via internet. This emerging media has become a dominant communication tool and it has been used as a communication channel in several events, especially “The Arab Spring” and BOSTON'S attack etc. In order to develop useful profiles of different cybercriminals, text mining techniques is an effective way to detect and predict criminal activities in microblog posts taking account the problems of data sparseness and semantic gap. The hashtags used on Twitter (e.g., #arabspring, #BostonAttack) contains outstanding indicators to detect events and trending topics especially to target and detect suspicious topics and eventual illegal events. Similarity approach is used in text analysis to detect suspicious posts in microblog publications. The evaluation of our proposed approach is done within real posts.