Deception detection in Arabic texts
The project Arabic Author Profiling for Cyber Security (ARAP) aims at preventing cyber-threats using Machine Learning. To this end, they monitor social media to early detect threatening messages and, in such a case, to profile the authors behind. Profiling potential terrorists from messages shared in social media may allow detecting communities whose aim is to undermine the security of others. One of this framework's main challenges ir recognizing false positives such as potential threatening messages that are actually deceptive, ironic, or humorous. This work focuses on the goal of detecting deceptive messages which are intentionally written trying to sound authentic. This task is performed on two different genres of Arabic texts: Twitter messages and news headlines.