Online Library

This policy brief explores new and innovative communications approaches to reduce the threat from all forms of violent extremism in Europe today based on a precise analysis of the strategic problem and corresponding strategic communications solutions deployable in response. In the context of an ever-evolving violent extremist landscape, including new and emerging forms of violent extremism, unfolding events in Israel and Gaza, and an accelerating digital environment – this policy brief looks at how governments can take a more strategic approach to preventing and countering violent extremism (P/CVE). Based on a strategic analysis of current events and an informed analysis of previous policy responses, this policy brief offers a new and practical approach to the use of strategic communications aimed at safeguarding all communities from all forms of extremist and violent threats, to turn back the tide of extremist influence for good.

“Twister,” developed by Miguel Freitas, is a social network platform centered around micro-blogging, much like Twitter. However, rather than relying on centralized servers owned and maintained by a single firm, Twister users operate a block chain combined with distributed hash table (DHT)–like and BitTorrent-like protocols to both make posts and send private messages, and also to receive entries from other users. Twister’s raison d’etre is that it offers a social networking platform that cannot be censored and cannot itself censor. The software does not record the Internet Protocol addresses users use to access the service, nor does it notify other users of an account’s online/offline status. Growing adoption of block chain services means that it is possible that the concept of decentralized social networks could become a norm. It is suggested in this article that block chain-based peer-to-peer social networks present challenges to the current counterextremist practices for content removal and censorship. While there are methods to disrupt usage of blockchainbased peer-to-peer services, these approaches may have the net harm of curtailing bona fide use of legal and novel technologies. Given this opportunity cost, non-transitory online violent extremist content may need to be tolerated.

This thesis describes the role that social media can play in showing how a terrorist organization can impact people’s conversation via Twitter. The two groups that this thesis focusses on are Boko Haram and Al-Shabaab. We present a new approach to how we can look into how terrorist organization can be analyzed and see what kind of impacts they may have over different cultures. The process used in researching and writing this thesis is we conducted literature search of the social media phenomenon and what social media can provide. We look to build on research by using the social media phenomenon to find what types of impacts terrorist organizations may have over cultures along with seeing how a terrorist event can have an impact over people on social media. This thesis hopes to expand on previous research on the academic uses for social media, as well as add to the expanding role that social media can be used for intelligence purposes.

This article contributes to the growing literature on extremist and terrorist online ecologies and approaches to snapshotting these. It opens by measuring Twitter’s differential disruption of so-called “Islamic State” versus other jihadi parties to the Syria conflict, showing that while Twitter became increasingly inhospitable to IS in 2017 and 2018, Hay’at Tahrir al-Sham and Ahrar al-Sham retained strong communities on the platform during the same period. An analysis of the same groups’ Twitter out-linking activity has the twofold purpose of determining the reach of groups’ content by quantifying the number of platforms it was available on and analyzing the nature and functionalities of the online spaces out-linked to.

Extremist online networks reportedly tend to use Twitter and other Social Networking Sites (SNS) in order to issue propaganda and recruitment statements. Traditional machine learning models may encounter problems when used in such a context, due to the peculiarities of microblogging sites and the manner in which these networks interact (both between themselves and with other networks). Moreover, state-of-the-art approaches have focused on non-transparent techniques that cannot be audited; so, despite the fact that they are top performing techniques, it is impossible to check if the models are actually fair. In this paper, we present a semi-supervised methodology that uses our Discriminatory Expressions algorithm for feature selection to detect expressions that are biased towards extremist content (Francisco and Castro 2020). With the help of human experts, the relevant expressions are filtered and used to retrieve further extremist content in order to iteratively provide a set of relevant and accurate expressions. These discriminatory expressions have been proved to produce less complex models that are easier to comprehend, and thus improve model transparency. In the following, we present close to 70 expressions that were discovered by using this method alongside the validation test of the algorithm in several different contexts.

From its start, the so-called Islamic State of Iraq and the Levant (ISIL/ISIS) has been successfully exploiting social media networks, most notoriously Twitter, to promote its propaganda and recruit new members, resulting in thousands of social media users adopting a pro-ISIS stance every year. Automatic identification of pro-ISIS users on social media has, thus, become the centre of interest for various governmental and research organisations. In this paper we propose a semantic graph-based approach for radicalisation detection on Twitter. Unlike previous works, which mainly rely on the lexical representation of the content published by Twitter users, our approach extracts and makes use of the underlying semantics of words exhibited by these users to identify their pro/anti-ISIS stances. Our results show that classifiers trained from semantic features outperform those trained from lexical, sentiment, topic and network features by 7.8% on average F1-measure.