By: Sabina Ciofu
Unlike any other terrorist organisation, the so-called Islamic State has consistently and efficiently made use of social media tools for self-promotion and recruitment. With an estimated 200,000 tweets a day for at least the last couple of years, it is by far the most aggressive social media offensive we have ever seen from a radical group. Western powers, often focused on traditional military and political responses to conflict and aggression, have initially reacted in a weak and fragmented manner to the wave of social media activity. It is only recently that concerted action has been taken by national and regional counter-terrorism authorities to respond to the threat of online radicalisation and recruitment. This has led to some decrease in the number and activity of English-speaking ISIS accounts, but it is still far from achieving the dismantling of their online networks.
That there is no coordinated military and political solution to the crisis in Syria and Iraq is getting more and more obvious by the day. The Western powers cooperate with the Arab countries and Sunni groups on one side, while Russia is closely working with Iran and the Shia militias on the other. With such a complex landscape of strategic interests, it is no surprise that coming up with a coherent approach to ending the war in Syria is proving a big mountain to climb. However, the latency to counteract ISIS’ charm offensive on social media – even when it targets Western citizens who have in significant numbers fallen for the Islamist rhetoric – is far more problematic.
Following a slow initial response, in recent times several meetings have taken place between Western governments and the largest American internet platforms in an attempt to cooperate in fighting ISIS propaganda online. The US government has, upon a number of occasions, asked the private sector for assistance, most recently enlisting Facebook, Instagram and Twitter to assist in the fight against terrorism. The French government has also contacted the US companies for support in removing online propaganda material, following the Charlie Hebdo attacks, while the UK Parliament had previously put blame on internet platforms, claiming that they are instrumental in spreading terrorist ideology. Far from a synchronized effort to find a common strategy to combat terrorism online, it oftentimes looks like a desperate cry for help from some governments, when they realise that what goes on online is strongly linked to national security threats. And this picture is further twisted by the complexity of the international law system, where tech companies are left with the decision of balancing out questions of freedom of expression, censorship and the difference between dissemination and promotion of online propaganda.
Using Big Data to draw patterns
What if one could turn one of the mightiest ISIS weapons into a liability? What if, by using big data analytics, one can look into the huge amount of content provided by ISIS-related websites, traditional media and social media accounts, to be able to draw relevant patterns? This may already be underway in highly classified intelligence programmes, where advanced algorithms may be used to track and determine potential terrorist activity. For example, in domestic law-enforcement domains, this is a model already being explored in some parts of the United States. By making use of mathematical and analytical techniques, police authorities are able to determine patterns that could lead to predicting criminal activity.
Apply that same technique for counter-terrorism and the value of big data analytics increases substantially. Having the technical ability to follow the data patterns, the footprints and the online records, looking into location, travel, profiles and messages of potential terrorists is a gold mine for national security authorities. It is easy to estimate that advanced big data analytics will outpace the computational ability of ISIS users to fake their identity and hide their location, as there are already a significant number of failings, even when instructed to hide their GPS location. There are situations where ISIS buildings and hotspots have been targeted by Western military forces, after users had accidentally pinpointed their location on social media. Ultimately, there is very little ISIS propagandists can do to completely hide online, if the organisation’s aim is to use the internet to spread the ideology, promote the mission and recruit new fighters.
As a general rule, recent research shows that social media analytics can be used for creating detailed profiles of potential terrorists and then looking for places where a high percentage of the local population matches the profile. Although this would seem a pretty straightforward approach, it has been shown that looking into the motivations and backgrounds of confirmed terrorists doesn’t necessarily lead to one single profile. However, sketching using big data analytics can define some widely-valid characteristics. For instance, ISIS recruits tend to be predominantly young and male and the ones originating in the EU and the US tend to come from a middle class background, with a high level of education. Drawing patterns out of huge amounts of available data has, however, obvious downsides, as it would only be a perfect profiling system if both the data and the algorithm were perfect. Therefore, privacy concerns are ultimately justified and any such governmental initiatives should duly take them into account.
Existing Big Data analytics projects
A massive data mining project by Qatar Computing Research Institute in Doha looked at social media data to figure out the origins of support for the Islamic State. Over a three month period, scientists looked into more than three million tweets, highlighting common patterns and attributes between pro and anti-ISIS messages. The algorithm was successful in “guessing” the sentiment in 87% of the cases, which is amazingly high for any big data project.
The Intelligence Advanced Research Projects Agency (IARPA) – the high-tech research arm of the U.S. intelligence community – is also focusing several programmes on using big data analytics to tackle some of the challenges ISIS has created online. For instance, they are currently focusing efforts on a facial recognition programme, based on imagery collected from available sources, be they high resolution cameras, or mobile phones, and even devices with less reliable resolution or lighting. Of course, the process is proving to be a difficult one, especially when algorithms are in charge of creating one face and recognising it from imagery in various angles, with varying lighting and quality.
Aside from its facial recognition programme, IARPA is also focusing on analytics results based on online video searching, through its code-named Aladdin programme. This implies designing new big data search methods for video content, that does not simply target tag words or user-generated content, but scans the video itself for elements that can describe what the movie is about. While terrorists may be clever enough not to tag videos where, for instance, they explain how to make explosives in view of a potential terrorist attack, their videos would be easily inspected and tracked down if a search method non-reliant on tags would be developed. Thus, while YouTube has been doing a great job in taking down such content almost in real time and while ISIS has moved these activities to the dark web to be less visible and thus less searchable, these movie samples may provide very valuable insight to national security experts, using advanced big data analytics to extract information from video material.
Moreover, the bigger the data, the better the big data analytics will be. Shared information for national security purposes has been something friendly governments have been focused on doing ever since 9/11. The precision and quality of information coming out of a big data project will always rely on the amount of data analysed, thus making it obviously necessary that countries cooperate in exchanging the information they have. The Dfuze system, for instance, is a database that allows such information exchange across multiple countries. National security experts can use the platform to access large amount of data shared by various actors and thus draw possible trends and patterns that can assist with prevention and preparation in view of potential terrorist attacks. There are already 40 countries using this product, which is an indication that such big data application can be very powerful tool in terrorism prevention and control.
This is not to say that big data, alone, can prevent terrorist attacks. Lone wolves will always be a difficult category to track, hunt and make sense of. This is to say, however, that big data can have a very significant input in creating and tracking the kind of patterns needed for effective intelligence gathering. It can also have a very important part to play in prediction, especially when it comes to planned, organised and coordinated terror attacks. Significant human reasoning and expertise will have to be attached to this, to distinguishing between real attacks and online bluff, between facts and intentionally deceptive mass-upload of messages. But that is the case for any intelligence method – human reasoning will continue to be at the centre of decision-making. And while it will always be the case that intelligence failures will be blamed for terrorist attacks – and hence also the failure of the people involved and the tools used – we simply don’t know how many terror plots have already been foiled and how many anti-ISIS military operations have already been successful with input from big data analytics.
Sabina Maria Ciofu is a first year MPhil/PhD candidate in Defence Studies, at King’s College London, where she explores the relationship between big data and US foreign policy. She is also a policy advisor in the European Parliament, working on digital economy, foreign affairs and trade issues. Sabina holds a BA in Classics from Cambridge and a MA in War Studies from King’s College London. @SabinaCiofu
Sabina Maria Ciofu
Sabina Maria Ciofu (@SabinaCiofu) is a second-year PhD candidate in Defence Studies at King’s College London, where she explores how machine learning is impacting foreign policy decision-making. She is also a policy advisor in the European Parliament working on the digital economy, foreign affairs and trade issues. Sabina holds a BA in Classics from Cambridge and an MA in War Studies from King’s College London.