Detection of terrorist threats on Twitter using SVM

Many people are exposed daily to different forms of terrorist threats on social networks, which make the control and detection of these contents paramount. We propose in this article, a system of detection of publications related to terrorism in the social network Twitter based on SVM (Simple Vector Machine). We established a 12-step process for analyzing, processing, and then detecting threatening tweets. We have developed 2 scenarios of use of this process, in the first scenario, we apply learning on 4000 tweets written in English, then 4000 tweets written in Arabic and finally 4000 tweets written in bilingual (in both languages). In the 2nd scenario, we apply learning on all the tweets at once (12000 tweets). These two scenarios allow us to compare the results obtained using SVM with results obtained using a syntactical approach.