An Ontology-Based Approach for Mining Radicalization Indicators from Online Messages

Abstract

Detecting radicalization on social networks is crucial to the fight against violent extremism and terrorism. In most cases, online radicalization has clear warning indicators that can be detected at the early stages of the radicalization process. In this paper, we focus on mining radicalization indicators from online messages by exploiting structured domain knowledge. More precisely, we propose an approach to automatically annotate social messages with concepts from a domain ontology. Annotations are then exploited within an inference phase to identify the messages exhibiting a radicalization indicator. We conducted a set of experiments on a sample extracted from a public dataset that contains radicalized individuals along with their social messages (i.e. Tweets). Obtained results show the effectiveness of our approach compared to a baseline method.