Online Library

We present a machine learning framework that leverages a mixture of metadata, network, and temporal features to detect extremist users, and predict content adopters and interaction reciprocity in social media. We exploit a unique dataset containing millions of tweets generated by more than 25 thousand users who have been manually identified, reported, and suspended by Twitter due to their involvement with extremist campaigns. We also leverage millions of tweets generated by a random sample of 25 thousand regular users who were exposed to, or consumed, extremist content. We carry out three forecasting tasks, (i) to detect extremist users, (ii) to estimate whether regular users will adopt extremist content, and finally (iii) to predict whether users will reciprocate contacts initiated by extremists. All forecasting tasks are set up in two scenarios: a post hoc (time independent) prediction task on aggregated data, and a simulated real-time prediction task. The performance of our framework is extremely promising, yielding in the different forecasting scenarios up to 93% AUC for extremist user detection, up to 80% AUC for content adoption prediction, and finally up to 72% AUC for interaction reciprocity forecasting. We conclude by providing a thorough feature analysis that helps determine which are the emerging signals that provide predictive power in different scenarios.

Recent terrorist attacks carried out on behalf of ISIS on American and European soil by lone wolf attackers or sleeper cells remind us of the importance of understanding the dynamics of radicalization mediated by social media communication channels. In this paper, we shed light on the social media activity of a group of twenty-five thousand users whose association with ISIS online radical propaganda has been manually verified. By using a computational tool known as dynamical activity-connectivity maps, based on network and temporal activity patterns, we investigate the dynamics of social influence within ISIS supporters. We finally quantify the effectiveness of ISIS propaganda by determining the adoption of extremist content in the general population and draw a parallel between radical propaganda and epidemics spreading, highlighting that information broadcasters and influential ISIS supporters generate highly-infectious cascades of information contagion. Our findings will help generate effective countermeasures to combat the group and other forms of online extremism.

Recent terrorist attacks carried out on behalf of ISIS on American and European soil by lone wolf attackers or sleeper cells remind us of the importance of understanding the dynamics of radicalization mediated by social media communication channels. In this paper, we shed light on the social media activity of a group of twenty-five thousand users whose association with ISIS online radical propaganda has been manually verified. By using a computational tool known as dynamic activity-connectivity maps, based on network and temporal activity patterns, we investigate the dynamics of social influence within ISIS supporters. We finally quantify the effectiveness of ISIS propaganda by determining the adoption of extremist content in the general population and draw a parallel between radical propaganda and epidemics spreading, highlighting that information broadcasters and influential ISIS supporters generate highly-infectious cascades of information contagion. Our findings will help generate effective countermeasures to combat the group and other forms of online extremism.

Much research aims to detect online radical content mainly using
radicalisation glossaries, i.e., by looking for terms and expressions associated with
religion, war, offensive language, etc. However, such crude methods are highly
inaccurate towards content that uses radicalisation terminology to simply report on
current events, to share harmless religious rhetoric, or even to counter extremism.
Language is complex and the context in which particular terms are used should not
be disregarded. In this paper, we propose an approach for building a representation
of the semantic context of the terms that are linked to radicalised rhetoric. We
use this approach to analyse over 114K tweets that contain radicalisation-terms
(around 17K posted by pro-ISIS users, and 97k posted by “general” Twitter users).
We report on how the contextual information differs for the same radicalisationterms
in the two datasets, which indicate that contextual semantics can help to
better discriminate radical content from content that only uses radical terminology.
The classifiers we built to test this hypothesis outperform those that disregard
contextual information.

VOX-Pol Blog post.

This report presents the analysis of a corpus of academic and grey literature relevant to a key
challenge facing our society. How can media and communications be used to counter identity-based
violence (IBV) or Violent Extremism (VE)?
Part I focuses on “Counter-Narratives”, looking at the evidence relating to strategic policy
communication strategies and counter-propaganda techniques. This reflects literature from
policymakers, think-tanks, and civil society initiatives rather than the academic literature base.
Key findings include the following:
• Current literature and policy concerned with countering propaganda is dominated by the language
of ‘counter-narratives’ but a common understanding of this relatively new lexicon has yet to emerge.
• There is little hard evidence that proves interaction with VE content leads to participation in VE
activities.
• The hypothesis that VE narratives or the real life threat of VE can be countered by an alternative set
of communications is an assumption that remains unproven.
These findings challenge claims that responding to propaganda strategies by firing back
with “counter-narratives” can be effective.
Part II looks at “Alternative Approaches” to the use of the media to counter violent extremism,
drawing on insights from the “media development” and “media assistance” sectors, and research
into whether mass media and new communication interventions can inhibit identity-based violence
in certain crisis situations. Key findings include the following:
• The theoretical foundations for these alternative approaches are supported by a stronger and more
established research base, drawn from the multi-disciplinary fields of development, peace building,
and social cohesion.
• Media projects have less impact if seen to be linked to a political agenda.
• A growing evidence base suggests that radio and television drama addressing issues of identity,
reconciliation and tolerance have a positive an impact on public attitudes and behaviour.
• Media assistance can ensure that local and domestic media can respond appropriately to VE
narratives.
• There is an emerging evidence base regarding the potential for rapid reaction media and
communication strategies in situations where there is a threat of IBV.
These findings suggest that alternative media strategies can help. But the trust and credibility of
information providers is crucial.
The final section “Reflections” concludes that the research landscape is fragmented and disconnected.
but suggests several professional/practitioner sectors and academic disciplines could shed light on
potentially effective media and communication CVE strategies.
More needs to be done to draw the threads together to learn lessons and to identify and prioritise
gaps in our knowledge and understanding.