Online Library

The Internet and, in particular, Online Social Networks have changed the way that terrorist and extremist groups can influence and radicalise individuals. Recent reports show that the mode of operation of these groups starts by exposing a wide audience to extremist material online, before migrating them to less open online platforms for further radicalization. Thus, identifying radical content online is crucial to limit the reach and spread of the extremist narrative. In this paper, our aim is to identify measures to automatically detect radical content in social media. We identify several signals, including textual, psychological and behavioural, that together allow for the classification of radical messages. Our contribution is threefold: (1) we analyze propaganda material published by extremist groups and create a contextual text-based model of radical content, (2) we build a model of psychological properties inferred from these material, and (3) we evaluate these models on Twitter to determine the extent to which it is possible to automatically identify online radical tweets. Our results show that radical users do exhibit distinguishable textual, psychological, and behavioural properties. We find that the psychological properties are among the most distinguishing features. Additionally, our results show that textual models using vector embedding features significantly improves the detection over TF-IDF features. We validate our approach on two experiments achieving high accuracy. Our findings can be utilized as signals for detecting online radicalization activities.

This paper explores various Social Network Analysis (SNA) techniques in order to identify a range of potentially ‘important’ members of Islamic Networks within Dark Web Forums. For this experiment, we conducted our investigation on five forums collected in previous work as part of the DarkWeb Forum portal and built upon the tool support created in our previous research in order to visualise and analyse the network. Whilst existing work attempts to identify these structures through state-of-the-art Computational Linguistic techniques, our work relies on the communication metadata alone. Our analysis involved first calculating a range of SNA metrics to better understand the group members, and then apply unsupervised learning in order to create clusters that would help classify the Dark Web Forums users into hierarchical clusters. In order to create our social networks, we investigated the effect of repeated author resolution and various weighting schemes on the ranking of forum members by creating four social networks per forum and evaluating the correlation of the top n users (for n = 10; 20; 30; 40; 50 and 100). Our results identified that varying the weighting schemes created more consistent ranking schemes than varying the repeated author resolution.

This article investigates Twitter data related to the kidnapping case of two German nationals in the southern region of the Philippines by the Abu Sayyaf Group (ASG). It explores perceptions of the ASG, along with associated organizations and sentiments indicated in the tweets together with statistically significant relationships. Findings revealed that: “Rebel” and “Militant” were the most frequently used labels for the ASG; a majority of the tweets contained sentiments that assess threats such as abduction and kidnapping of hostages; and almost half contained words that indicate negotiation or concession to the demands of the captors. Logistic regression analyses on “Rebel” and “Islamist” revealed positive coefficients for these sentiments used as predictors. This meant that people who assessed threats and expressed sentiments that responders should concede to the captors’ demands were more likely to use the “Rebel” or “Islamist” labels. Rather than the two longstanding dominant narratives of the ASG as terrorists and criminals, the emerging rebel and militant labels suggest a more domestically and politically sensitive Twitter commentary than is represented in the work of the Al-Qaeda-centric paradigm exponents. These findings, along with the complex associated political and policy contexts and implications, are discussed in this article.

VOX-Pol Blog Post.

The rise of the Internet has dramatically increased the degree to which youth may be exposed to online hate content, and simplified the process of sharing this content with others. Viewing messages that contain hate speech or language vilifying others can increase an individual’s risk of radicalization to extremist views and the acceptance of violent ideologies. Researchers have begun to explore the risk factors for exposure to such content, with prior studies demonstrating a relationship between low self-control and online activities being important correlates. Few studies have utilized youth samples to assess these relationships, or explored the voluntary consumption and sharing of content. This study attempts to address this gap in the literature using self-report responses provided by a sample of 1,193 youths in South Australia. A series of quantitative models are estimated assessing the relationships between self-control, opportunities to view content using both on and off-line measures, and four dependent variables related to exposure to or sharing of hate content. The implications of this analysis for our understanding of the utility of criminological theory to radicalization and countering violent extremism are discussed in detail.