ADA 2019 - Projet
Terrorism perceived through western news
Abstract
Photo credit: [_0_h_m_]

Abstract

Since September 11, 2001, terrorism has been in the headlines around the globe. A general fear reappears every few years as events labelled as "terrorist attacks" occur. Certain parties have gained in popularity in the past few years, often using the anxiety of "dangerous terrorists" as a pretext to defend anti-immigrant policies. Have terrorist attacks increased in the past years as we are led to believe? Or has the rate been more or less constant? Who commits these attacks and what population is the real target? News coverage plays a big role in how we perceive these events, some events are debated and discussed during weeks, while others are never mentioned. We want to explore the rate and distribution of these attacks over the world, and dive into their news coverage in the western hemisphere. What themes are covered in parallel to the articles linked to terrorism? We will investigate what groups are considered “active terrorist” groups around the globe and the magnitude of their attacks and their damage.

Research Questions

  • Have the terrorist attack rates and locations, changed over the past 50 years?
  • Where do terrorist attacks occur?
  • Who commits terrorist attacks and with what target ?
  • How are is global terrorism depicted in some selected popular western newpapers?

Sources

Global Terrorism Database (GTD) is a collection of facts about terrorist attacks from 1970-2017. The facts cover location, date, targets, and who claimed the attack and a very large variety of other very specific metrics. GTD was built by a multidisciplinary team of University of Maryland faculty members and contains a large variety of structured variables characterizing each attack’s location, tactics and weapons, targets, perpetrators, casualties and consequences and general information such as definitional critera and links between coordinated attacks. Event from 1993 were scraped from Wikipedia, as they were missing in the initial dataset.
All the news is a collection of news coverage from CNN, New York Times, Breitbart, Fox news, the Guardian. New York Times articles was used to complete the dataset of articles collected from “all the news” as this was a rather narrow dataset. For the article selection concerning terrorism, a score was built by scanning the general bodies for an overlap with the dictionary of “terrorism” in spacy. They were then also scored on various other topics, and if terrorism was among the top 3 scores, the article was retained. Articles with general low scores were also removed.

Datastory

To be (a terrorist) or not to be? GTD Definition of Terrorism and Inclusion Criteria. The GTD defines a terrorist attack as the threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation. In practice this means in order to consider an incident for inclusion in the GTD, all three of the following attributes must be present:
  • The incident must be intentional – the result of a conscious calculation on the part of a perpetrator.
  • The incident must entail some level of violence or immediate threat of violence -including property violence, as well as violence against people.
  • The perpetrators of the incidents must be sub-national actors.

  • The database does not include acts of state terrorism. In addition, at least two of the following three criteria must be present for an incident to be included in the GTD:
  • criterion1: The act must be aimed at attaining a political, economic, religious, or social goal. In terms of economic goals, the exclusive pursuit of profit does not satisfy this criterion. It must involve the pursuit of more profound, systemic economic change.
  • criterion2: There must be evidence of an intention to coerce, intimidate, or convey some other message to a larger audience (or audiences) than the immediate victims. It is the act taken as a totality that is considered, irrespective if every individual involved in carrying out the act was aware of this intention. As long as any of the planners or decision-makers behind the attack intended to coerce, intimidate or publicize, the intentionality criterion is met.
  • criterion3: The action must be outside the context of legitimate warfare activities. That is, the act must be outside the parameters permitted by international humanitarian law, insofar as it targets non-combatants.
  • Have the terrorist attack rates and locations, changed over the past 50 years?

    Evolution of number of terroror attacks in the last 50 years
    This graph generally shows an increase of number of attacks over the past 50 years. We can also see that the data scraped in attempt to complete the events from 1993 is not comparable to the number of events recorded for the other years. Specific norms for data collection are absent before 1997, therefore, we will focus on the years 1997-2017 for the following analysis. The trend in the early 2000’s was relatively low, which may come as a surprise as this is when terrorism became a hot topic. Between 2013-2017, the number of attacks have decreased. This could partially be due to change in legislative regulations. But is this a global trend?


    Casualties per year

    The number of casualties seem to follow the same trend as the number of attacks, except for 2001, where we have an important peak in deaths and wounded victims. This peak relates to the 9/11 terror attacks, we can see that in terms of damage either related to human casualties or property, this was one of the major events in the decennia. However, in more recent events, the total number of casualties almost doubled in comparison to the memorable event of 9/11.
    The plot on the right depicts the magnitude of violence of the terror attacks by normalizing the amount of casualties to the number of attacks. The peaks clearly depict the single deadliest terrorist acts. Interestingly, after 2008 the number of casualties per attack seemed to decrease. We could hypothesize that this reflects the consequences of counter-terrorism or that the terrorist’s tactics have evolved. Very large scale attacks such as 9/11 are probably not as easy to reproduce anymore due to the increased security in the air traffic. On top of that, most countries that are not actively involved in conflicts have given more importance to foreign defence.


    Property damage per year

    Concerning property damage, we can clearly see a peak in 2001, representing the destruction of the World Trade Center in New York and its surroundings. The destruction of the Twin Towers, in terms of physical goods, is the largest damage that any terror group has achieved so far. When combining the material loss and huge human death toll, 9/11 still is the most traumatic event of these last 20 years. We can note that in 2014 and 2016, the normalised property damage is lower than the total count. This implies that the number of attacks in those years is the reason for the peak in damage, and that it is not necessarily due to the severity of the attacks. Concerning the other years, normalisation of the property damage according to the number of attacks does not result in a significantly different distribution.

    Where do terroror attacks occur?


    Distribution of terroror attacks in the world

    Since 2004, the country with the highest number of casualties is clearly Iraq, rapidly followed by Afghanistan and Pakistan. Middle East is the theatre of a large variety of complex conflicts, in particular during these last few years. Religion, ethnic, political and ideological conflicts are still ongoing in the Middle East. It may be interesting to keep in mind that USA was involved in most of these conflicts and wars... We can see that some African countries are also in very instable political situations, such as Nigeria and Somalia, from 2011 onwards.

    Who commits terroror attacks and with what target ?


    Most active groups of terrorists in the world

    We chose to plot the 30 most active terrorist groups in terms of number of attacks.
    As mentioned previously, the groups that are the most active are located in the Middle East with the Talibans in Afghanistan and Pakistan, ISIL in Iraq and other countries in the Levant region, Al-Shabaab in Somalia, Boko Haram in Nigeria and the neighbouring countries. In the recent years, the most active terrorist organisations are radical Islamist groups.
    There are a few groups that are being very active, and have committed several thousands of terror attacks. These groups, are large structured organisations, they have the fire power to rivalise with local order forces.


    Most targeted types by active groups of terrorists

    We can see that each group has a slightly different distribution of targets. Civilians are the primary targets of most groups, this being consistent with most groups strategic practices. Nevertheless, the targets of the most active groups are mainly directed towards governmental institutions. This could be interpreted as a protest against the established government, which they are trying to disrupt in order to gain in power. Instability in a countries order gives leanway to these large organisations to spread their message and ideologies.


    Targets

    If we dive into the individual types of targets, many subcategories can be investigated. Interestingly, we can see that civilians are the primary victims of terrorists followed by the police, military and government. In all categories, primary targets still are individual peoples rather than properties, this indicates that terrorists aim not to disrupt pure communication or material supply of the countries they target, which would be one of the most classical war strategies to weaken governments and countries. We hypothesize that the strategy of targeting individuals has the aim to shock and traumatize populations as the aim is not necessarily to target a specific subtype of people, but generate as many casualties as possible. The fact that “random” civilians are likely to be the general targets induces fear among the general population. Fear is their way of communicating ideological views of the groups in question and is what gives terrorism its social and ultimately political punch.
    Main forces that act against terrorists are also largely targeted. The Police and Military are classes of people that may be the most directly related to counter-terrorism and prevent a country from sinking into chaos. They are important forces in a country’s organisation.


    Favorite weapons of active groups of terrorists in the world

    We would have expected each group to use different weapons depending on how easy the access to some resources is. But even here, the most popular weapons for most groups are explosives and firearms. This seems to make sense as we saw that targeting civilians at random and wanting to induce a large amount of damage is key. It is possible that the black market is responsible for the weapons to reach the groups in areas where resources are limited. Palestinian Extremists surprisingly have to resort to use melees as weapons, this might mean that the blockage of the Gaza strip might actually prevent firearm traficants to reach Palestine and they resort to less sophisticated weapons.
    https://en.wikipedia.org/wiki/Blockade_of_the_Gaza_Strip


    Repartitions of the 100 most active groups of terrorists

    This figure represents the repartition of the 100 most active groups per continent in a logarithmic scale. As we can see, the most active groups are in Asia, indeed that confirms what was seen previously in the Middle-East. Africa is also one of the continents where groups are most active, again this is consistent with was we saw previously. Thanks to this graph and the one of most active terrorist group in the world will enable us to generate the following map of the predominant groups per continent. This also allows us to establish that the magnitude of the attacks are not comparable among the different continents as the scale of activity is so different. The top “terrorist group” in Asia has committed over 7000 attacks, while the top North American group committed only around 60.


    Groups per continent


    The names of the different groups active on a certain continent are represented, have their size increasing proportionally to the contribution they bring to the total activity of the terrorist attacks of their continent. The size is only relevant within the own continent, and cannot be compared between continents! It might be interesting to note that terror groups within North America or Europe are not necessarily what one generally might imagine when one thinks of typical “terrorism”.

    How is global terrorism depicted in some selected popular western newspapers?


    Themes


    Using the articles database, we attributed to each article a terrorism score based on word content: words related to the terrorism semantic field. We also selected a handful of other topics and built a word map using these topics. We can now see the most represented topics in American articles appearing in relation to articles treating terrorism. Government, war and politics are topics relatively important in the articles. This could indicate that in the articles selected, the focus on the context of conflicts the general background/ political motives and situation is elaborated. They do not seem to focus on sentimental impacts of terror but rather on more factual information. However we may need further analysis of much more data to assert this.


    Themes per continents

    (AF: Africa, AQ: antarctica, AS: Asia, EU: Europe, NA: North America, OC: Oceania, SA: South America, the scales for the various years are not the same!) To take another look at the topics and themes of articles about terrorism, we built a radar plot with the scores of articles on a variety of themes to depict the angle with which terrorism appears in the different continents. We still see the same trend as in the word map, but some differences seem to emerge among the continents. Politics and government seem more important for articles about terrorism in Europe and North America. Overall, the predominant themes emerging from this plot seem to be : government, crime and war. It is difficult to observe general temporal trends, but we can see that in more recent years, from 2010 to 2017, war is an important topic linked with Asia.


    Article Coverage

    To be able to assess the article coverage, we attributed to each terror attack, the articles that cover it by choosing the articles dates and countries that are referred in the article. To further filter out unrelated articles, we computed the word similarity between the article headline and terror attack summaries and removed articles that had a zero similarity value. Here, we plotted in grey the casualtie ratios per country for a selection of countries. The casualties ratio was defined simply by the total number of casualties in a country divided by the total of all countries and similarly with the number of published articles related to terror attacks.
    We can clearly see a typical case of filter bubble, where sharing media mostly cover terror attacks that concern the United States or touch countries culturally close to the United States. The unbalance between the damage due to terror attacks and their coverage indicates clearly that some countries are more covered than others, this imbalance is clear and quite logical for the United States as the journals are analyzing here are all American. We can also see that even though there were a very few events in France and Germany these were covered quite well in comparison. African countries were poorly covered even though they are sadly the a significant portion of local terror events. Interestingly Iraq, Afghanistan and Pakistan media coverage was less skewed than we expected. This can be explained by the fact that the U.S. army was and still is tightly involved in wars in Iraq, Afghanistan and Pakistan. So naturally the media and the public is concerned and interested in the fate of their soldiers. Another hypothesis is that the amount of articles is related to the amount of casualties, the casualties in the middle east being relatively high, it makes sense that the media coverage would follow. One extrapolation of this relatively high coverage of the middle east could interpreted as being used as political propaganda to sustain a certain fear among the American population for these countries and what is occuring there.


    Article length

    This plot represents the average word count in articles concerning terror attacks in a selection of countries. First of all,we can see that the distribution is relatively unstable and that no trend can be directly be interpreted. We can observe some higher spikes within Somalia and Afghanistan. we suppose that for terrorism in Somalia only a few articles were published but they were most likely detailed, resulting in rather large text count.
    The War in Somalia started in 2006, so maybe it’s not that surprising to find few references to this country in the article database before this date. There is the possibility that the metric used is not as robust as we would have liked, so that it will be slightly difficult to draw conclusions on these kind of plots.


    Most active groups in Iraq

    When taking a first look at the plot, the most striking thing on this plot begins around 2012 and it is the increasing influence of ISIL becoming by far the most active group in Iraq.
    In fact most of the groups listed here are sub-factions of the same organization or allied ones. Islamic State of Iraq is the old denomination of the Islamic State of Iraq and the Levant. Tawhid and Jihad was the predecessor to Al-Qaida in Iraq and also the Islamic State of Iraq and the Levant. Terrorist organisations are clearly not static and their structure and hierarchy changes in time. Even if the terrorist groups are depicted as different ones on the plot, all these organizations are tightly linked and have undergone deep restructurations. Overall, with this in mind we can say see that these last few years, ISIL seemed to get more aggressive.


    Themes scores for Iraq or United States of America

    The evident peaks around 2016 are probably due to the highly unbalanced representation over time of articles within our dataset. This makes it hard to conclude anything about the evolution of the themes during the years. But it is interesting to note that the USA has far superior scores for those themes than Iraq, meaning that government, war and crime are more treated within articles linked within the United States.

    Conclusion and Limitations

    Our main limitation concerns the articles dataset. Indeed, most articles in the database were published between 2017, making our dataset very unbalanced. This has a great impact on news coverage in time, causing any plot with analysis in time of coverage of terror attacks coverage before 2015 to have few data points. Unfortunately, the plots are therefore not as representative as we would have wished. Another concern about this dataset is about the selection of the articles that may concern terrorism. As the dictionary used to chose the themes that the articles address is composed by terrorism related terms, it is sometimes possible that the accuracy of selection is not ideal. It would be interesting to find a tool that could better discriminate between the articles. A classifier would have be a useful tool, but unfortunately we didn’t have article’s themes labeled, and labeling the articles by ourselves would have been an impossible task. Furthermore the definition of Terrorism is a very complex one as mentioned in the beginning. Labeling the dataset would have been biased as to our personal perception of which event is considered as terrorism. Another problem encountered was during the extraction of countries from the content of the articles, it wasn’t possible for us to know if the countries extracted are the country where the attack took place or if they were just mentioned in the text for other reasons. General semantic analysis would have been required to extract this information.