Dutch Refugee Crisis on Twitter

Issue Analysis and Over Time Animation

Team Members

Project Proposal by:

• Arnout Ponsioen
• Gijs Van Beek

Facilitators:

• Richard Rogers
• Mathieu Jacomy

Student Members:

• Ysabel Camus
• Heleen Gorris
• Hatixhe Raba
• Vincent Sondermeijer

Project Presentation

Here are the slides that were presented on January 15, 2016 at the DMI Winter School:

• Dutch_Refugee_Crisis_slides.pdf: Dutch refugee 'crisis' over time on Twitter

Contents

Introduction

Starting in 2015, Europe was overwhelmed by an unprecedented number of refugees and migrants from war-torn Middle Eastern countries, especially from Syria. As a result, a number of EU countries began searching for new policy measures to prevent more people from fleeing the Middle East and to extend support for those already within European territory (Ponsioen and Van Beek).

As in other European countries, the influx of refugees in the Netherlands (see figure 1) led to heated political debates, expressions of anger and dissent, and a number of violent outbursts from the public. This issue is reflected online via thousands of interactions on various social media platforms (ibid.). This project focuses on Twitter as a platform facilitating the refugee issue.

fig. 1: Quantitative influx of refugees in the Netherlands

The following paper covers findings and new insights regarding the main voices in the debate, the majority of related content spreading around the platform, and the user networks surrounding the issue. These findings were gathered using tools from the Digital Methods Initiative (DMI), and visualized into static graphs and tables using Gephi and Tableau Public. However, the research project was also held simultaneously along the development of a new software tool that could visualize the data with over time animation.

The team began the project with an end-goal in mind: to gather data that could help influence a change of policy in the Netherlands. “A deeper understanding of the evolution of online issue networks is quite crucial for governments and NGO’s that needs to build appropriate participatory communicative network approaches aimed at building more trust, understanding, and respect” (Ponsioen and Beek, van).

Initial Data Sets

The data set used for answering the research questions consists of all tweets in the Dutch language from July to December 2015 using the following hashtags: #vluchtelingen, #asielzoeker, #azc, #noodopvang, #coa, and #migranten. In total, we worked with 245,364 tweets and 46,738 unique users. 126,468 of these tweets contained links or URLs, most of which led to relevant media that spur the debate online (see fig. 2).

Search query: #*vluchtelingen* OR #asielzoeker* OR #azc OR #noodopvang* OR #coa #OR #migranten

fig. 2: Percentage of Tweets containing links and no links

Research Questions

The project has four research questions addressing the challenge of gathering substantial knowledge from the data sets.

1. To what extent is the refugee issue a crisis?
2. Is it artificially sustained by an (organised) group of Twitter users of a particular (political) persuasion?
3. How to account for the decline of the activity in the Twitter discussion? Does Twitter ‘only’ follow media events?
4. Is the decline attributed to the end of the crisis or the end of the organisation of voice? Were particular voices silenced? Was there a conclusion to a debate?

Methodology

To retrieve the subsets of data, the team used the DMI Twitter Capturing and Analysis Toolset (DMI-TCAT). In order to analyse the content, we extracted tweet statistics such as “URL Frequency”, “Hashtag Frequency”, “Identical Tweets Frequency” and “Media Frequency”. The “URL Frequency” allows one to see which sources are referenced the most. We also extracted “Hashtag Frequency” to find out which hashtags are most often associated with our object of study. The “Identical Tweets Frequency” extraction contains tweets and the number of times they have been retweeted. This can give one an overview of the most “popular” content. Lastly, the “Media Frequency” extraction shows the media URLs and the number of times they have been used on twitter (DMI-TCAT). This gives one an overview of the most popular media.

In order to grasp a better understanding of the dominant voices leading the debate, the researchers also gathered activity metrics from DMI-TCAT, such as “User Stats (overall)”, “User Stats Individual”, “User Visibility”, “User Activity (tweet frequency)”, and “User Visibility+Activity”. We gathered the “User stats (overall)” to view the number of tweets, URLs, followers, and friends per user. The “User Stats Individual” then ranks the users according to the number of tweets, followers and friends, and how many times they appear in the data set. These two extractions give one a better feel of the users within the data set. Next, to determine the more influential voices, we used “User visibility (mention frequency)” to show which usernames were mentioned most often by others. Furthermore, to see whether the whole discussion is actually dominated by a certain group of people, we also extracted “User activity (tweet frequency)” to examine the amount of tweets posted by certain usernames. Lastly, to see whether the users mentioned are also those who tweet a lot, we also gathered “User visibility+activity (tweet+mention frequency)” to get the list of usernames alongside both the tweet and mention counts. All these extractions were made to better define the actor composition within the data set, and to arrive at a better understanding of the most important users and voices (ibid.).

Finally, to perform the network analysis we extracted the “Social Graph by Mentions”, “Bipartite Hashtag-user Graph”, and the “Co-hashtag Graph”. According to the tool, the “Social Graph by Mentions” extraction shows the interactions between users and can be used to analyze patterns in communication, to find communities and to categorize users. The “Bipartite Hashtag-user Graph” feature produces a bipartite graph that allowed us to explore the relationship between users and the hashtags they use. It allows one to find and analyze which users group are active regarding certain topics. Lastly, the “Co-hashtag Graph” shows which hashtags most often appear together, allowing us to explore the relations between hashtags and to determine sub-issues within the debate (ibid.). Unfortunately, among these extractions we were only able to visualize and analyze “Social Graph by Mentions”.

Each extraction was filtered to a threshold of 2, 5 or 20 according to what would best suit the data. For each extraction that garnered relevant findings, the team used Tableau Public and Gephi to visualize it into various tables, charts and graphs. All in all, one can cluster all the data retrieved according to what we wanted to analyse, which in this case include the content, the actors or users, and the network.

To expound on our approach towards the data set, we wished to address the need for ‘critical analytics’. Hence, we referred to the following five metrics given during the DMI Winter School 2016 (Rogers):

1. Dominant voice
2. Concern
3. Commitment
4. Positioning
5. Alignment

Using these metrics, we realized that we had to delve into the actors, that is, who is concerned and who are the dominant voices. We also wanted to look into the keywords to see the arguments or topics that circle around the debate. We observed the commitment level, which is shown by how long actors or keywords remained prominent within the time period.

Applying these metrics directly to the task at hand and the available tools, Ponsioen and van Beek clustered the extractions needed to meet each metric in the following manner:

1. Dominant voice and Concern

   1. Number of actors

   2. Number of tweets (per actor)

   3. Number of retweets and mentions (per actor)

      1. Indegree (authority)

      2. Outdegree (engaging)

      3. Eigencentrality

2. Positioning and Alignment

   1. Number of keywords/hashtags used

   2. Number of keywords/hashtags (per actor)

   3. Number of co-words used (per actor)

      1. Degree (undirected)

3. Commitment: spread of activity over time (how long is an actor or keyword active/dominant).

Furthermore, the team tried to bridge together actors and keywords and to identify tensions or abrupt changes in the data. To observe changes over time, we calculated statistics for specific time periods and visualized them in simple charts and graphs.

Findings

Timeline of Twitter Activity from July to December (2015)

By observing fig. 3 one can pinpoint noticeable spikes in Twitter activity revolving around the refugee issue during the last six months of 2015. The highest peaks in activity occurred on the 4th and 23rd of September, the 7th and 15th of October and finally on the 16th of December.

fig. 3: Timeline of Twitter Activity (from July 1, 2015 to December 22, 2015)

Those dates more or less correspond with significant news events that spur conversations related to the refugee issue online (see fig. 4).

fig. 4: Timeline of Twitter Activity with corresponding images

The spike in Twitter activity on the 4th of September can be attributed to the proliferation of Aylan Kurdi’s photo on official news websites, which was quickly amplified by social media. The photo depicts the body of a 3-year old boy that washed ashore on a Greek island. He drowned alongside his mother and 5-year old brother in their attempt to flee from war-torn Syria. This is reflective of Twitter activity even outside of the Netherlands, where the refugee issue was also being discussed globally, especially in the European area. The conversations surrounding the image mostly express sympathy for the hundreds of refugees who lost their lives while attempting to cross dangerous waters to flee from war. Hashtags like #HumanityWashedAshore, #KiyiyaVuranInsanlikand (Turkish for “humanity washed ashore”) and #DrownedSyrianBoy accompanied the image.

For the 23rd of September, the spike coincides with a meeting held by the European Council attended by the heads of state and governments in the European Union. These EU leaders met in Brussels to discuss the flow of migrants coming to Europe and how to “establish a credible European migration policy” (European Council). The meeting determined the priorities for immediate application and how to respond to long-term migratory challenges. It also concluded in calling for “renewed diplomatic efforts to solve the crisis in Syria” and to “ensure the formation of a government of national unity in Libya” (ibid.).

The highest peak on the Twitter Activity Timeline occurred on the 7th of October. During this time period, violent outbursts happened in Oranje. Oranje is a small rural town in the Netherlands originally inhabited by 130 people. A year prior, 700 asylum seekers from places like Syria, Sudan, and Eritrea were housed in a disused vacation camp in the town (Corder). When junior Justice Minister Klaas Dijkhoff announced that they would be sending 700 more, villagers reacted in violent opposition (ibid.). This was shortly followed by a series of similar protests in other towns like Beverwaard and Steenbergen, causing another spike in Twitter activity around the 15th of October.

After a period of relatively low activity related to the refugee discussion, the chart indicates one last jump in Twitter activity. The final spike during the period under observation stemmed from riots that broke out in the Dutch town of Geldermalsen. During the town council’s meeting to discuss plans of building a new centre to house 1,500 migrants, protesters caused a riot by tearing down fences and throwing fireworks at the police (BBC).

Thus, it is evident that the refugee discussion on Twitter is fueled and heavily influenced by significant media events. Each peak in Twitter activity almost directly corresponds to a parallel news event, leading the researchers to believe that each jump is an online reaction to real world incidents.

Actor Composition Analysis

The main online actors were analyzed according to the user activity, or how often they tweeted, and their visibility, that is how often they were mentioned by other users. The following analyses provides the users that come up in the top ten results of each categpry. After determining the top ten, the research will go into a more detailed account of each relevant user and their role in the discussion. This can result in an effective combination of quantitative and qualitative approaches towards analysing significant Twitter users.

To start off, the user activity, which is observed by their tweet frequency, can be visualized in the following graph (see fig. 5). Here, the most active users are ranked according to the frequency of their tweets.

fig. 5: Top Users Ranked by Frequency of Tweet Activity (users that tweeted the most)

The most active users were then determined, and further visualized into the graph below (see fig. 6), where the top 10 most active users’ activity were spread out over time. Upon categorizing these top users according to their stance in the argument, we found that nine out of the top ten most active users are extreme right wing. These user accounts strongly express that they are against receiving the influx of refugees or migrants into Dutch society.

On fig. 6, it is also evident that sometime in the middle of October 2015, most of the voices quieted down and ceased to continue the discussion. That is, except for @AZC_Alert_DB whose activity increased during this period of silence. This is interesting because the account was virtually mute at the time when most voices were tweeting about the issue in September and early October. This is where the findings first indicate a certain artificiality about the conversation regarding the refugee issue. This one account evidently exerted a lot of energy into sustaining the discussion after all the other participants had ceased to talk, and at a constant, high-frequency pace nonetheless.

fig. 6: User Activity (Tweet Frequency), spread out over time

As you can see from fig. 6, the most active user account is actually @Azc_Alert_DB, which is a local citizen platform for the city of Den Bosch. The account’s frequent activity is attributed to all the updates they give regarding the policy addressing asylum seekers. They also send out warnings to people from certain areas whenever plans for a new refugee shelter comes up, so that they can protest the plans. This account is considered extreme right wing.

fig. 7: Twitter profiles of @AZC_Alert_DB and @PeterHandrickse

The second most active user is @peterhandrickse, who is an anonymous individual strongly opposing refugees and everything that is not Dutch. He calls himself “provocative”. He tweets frequently about the issue, although his following is not very significant (only 58).

fig. 8: Twitter profiles of @roepme and @EindhovenHolly

The third most active account belongs to @roepme. His Twitter description reads, “Atheïst judeo-christian hindu buddhist internethero suffering from islam allergy also known as infidel Gematigd dyslectiticus niet te min vermoedelijk D(i)slect.” Here, he states that he is “allergic” to Islam as is evident from his tweets, which strongly oppose refugees.

The fourth most active account is @eindhovenholly. In relation to the refugee issue, this account mostly retweets rather than creates original tweets. He or she often retweets Geert Wilders, a Dutch extreme right-wing politician, right-wing media and individuals opposing refugees.

fig. 9: Twitter profiles of @ArmandVervaeck and @AZC_Alert

The researchers found the fifth account very intriguing. This belongs to @ArmandVervaeck, an individual with no specific authority on the refugee issue. He is an earthquake specialist residing in Belgium. However, his account frequently came up in our data because of his use of the Flemish language. He tweets anti-refugee sentiments regarding their entry into Belgium.

The sixth most active account belongs to @AZC_Alert, the “mother” account of @AZC_Alert_DB. Their twitter description reads, “Landelijk burgerplatform, strijdend voor meer inspraak en een humaan asielbeleid waarbij de nadruk op opvang in regio ligt.” Roughly translated, this is “Nationwide citizen platform, fighting for more participation and a humane asylum policy per region.” The account positions itself as a “citizen platform”, rather than a media platform. They are clamoring for the asylum policy to be authored or implemented on a regional basis, rather than on a national level. They are using Twitter to mobilize people against refugees. Like their Twitter account for Den Bosch, they warn followers whenever there are plans for new refugee shelters in a specific area. In general, they stimulate citizens to speak out whenever they are not satisfied with the government.

fig. 10: Twitter profiles of @Kidefrian and @Hannesz1956

The seventh account that tweets most frequently is @Kidefrian, an individual who is against both Islam and assimilating refugees into Dutch society. In his Twitter description he uses the hashtag #kominverzet, which is a call for people to join the “resistance”. This hashtag is commonly used by right-wing people to protest the presence of Islam in the Netherlands.

The eighth account is called @Hannesz1956, another individual against refugees and Islam. His Twitter biography reads, “Hannes de Zanger. Leerde Agitprop schrijven op Journalistenschooltje Utrecht. Zanger, muzikant & acteur. ANTI islam want DEMOCRAAT. PRO NL & Israël. Patriot.” In English, he describes himself as a singer, musician and an actor. One can also gather that he teaches journalistic writing in Utrecht. He further describes himself as anti-Islam and anti-Democrat, but pro-Netherlands and Israel. He concludes that he is a Patriot. However, he seems to have deleted his account rendering him unsearchable, though his tweets are still visible.

fig. 11: Twitter profiles of @AdriaanBeenen and @BastaTV

The ninth most active account is @AdriaanBeenen, another right-wing individual who describes himself as a practical philosopher. He is against refugees coming into Holland and is also very much against the European Union.

At the bottom of the top ten most active users is the only account that shares left-wing views. This is @BastaTV, an educational platform by the local broadcasting company AT5. Their description reads, “Basta is het educatieve platform van AT5 voor groep 7 en 8. Deze maand: Amsterdam Helpt! (Vluchtelingenhulp) Ga naar http://www.bastaschooltv.nl #bastatv”. In October, their monthly theme was “helping refugees”. This is the lone left-wing voice among the top ten. Hence, in terms of user activity or tweet frequency the extreme-right wing voice dominates the conversation.

Moving on to user visibility, which is a combination of how often a user tweets and is mentioned by others, the significant findings are illustrated in the following table (see fig. x). The right-most column indicates the number of times the account was retweeted during the time period.

fig. 12: User Visibility (Tweet+Mention Frequency) Table

The most notable account in terms of visibility is @ArmandVervaeck. Considering his background as an earthquake specialist, the team initially questioned his legitimacy as a source within the refugee issue. We are left to wonder how and when his legitimacy was built up. This was a challenge on our part, since it required retracing the history of his engagement with the topic. So far, he becomes legitimate in the sense that he is mentioned and retweeted very often within the debate. @AZC_Alert also comes up in this list, proving that it is both active and visible within the debate.

fig. 13: Twitter Profiles of @TerreurMonitor and @PeterAnshof

It is also interesting that the account @TerreurMonitor (depicted above) is frequently mentioned or retweeted in relation to refugees. The account exists to monitor jihad and terrorism threats, as seen in their twitter biography; “Jihad en Terrorisme Dreigingsmonitor Contact: info@terrorismemonitorREMOVE_ME.nl”. Translated, this says “Jihad and Terrorism Threat Monitor Contact: info@terrorismemonitorREMOVE_ME.nl”. This is one piece of evidence suggesting that the right-wing uses terrorism, or fear of it, as an argument to steer audiences against the presence of refugees.

On the other hand, @PeterAnshof is another extreme-right wing voice with a surprisingly large amount of followers. He is associated with the Netherlands’ Security and Investigations department. Via Twitter, he comments about politicians’ statements about refugees in real-time, and noticeably refers to @AZC_alert, @telegraaf and @geertwildersPVV, who are also prominent right-wing voices on Twitter.

fig. 14: Twitter Profiles of @telegraaf and @RodeKruis

@Telegraaf is another account which is high in visibility. It is a large, traditional newspaper that expresses right-wing views. Their twitter description says, “Officiële account van de krant met het nieuws van ons land | dagelijks het laatste nieuws | binnenland | buitenland | financieel | privé | sport | DM voor tips”. Translated this simply says that is the “Official account of the newspaper with news about our country | daily latest news | domestic | abroad | financial | private | sport | DM for tips”.

Finally, another left-wing voice enters the network of actors within the debate. The official account of the Dutch Red Cross is also a visible actor, sixth on the list. The description reads, “Het Rode Kruis is de hulporganisatie op het gebied van noodhulp, zelfredzaamheid en EHBO. Heb je vragen aan ons? Wij staan hier voor je klaar.”, which translates “The Red Cross's relief organization in the field of emergency response, first aid and self-reliance. Do you have questions for us? We are here waiting for you.”

fig. 15: Twitter Profiles of @geertwilderpvv and @volkskrant

A personality that was expected to come up in this research was Geert Wilders. He is the 7th most visible Twitter personality participating in the refugee issue. His account, @geertwilderspvv, holds the description “Voorzitter Tweede Kamerfractie Partij voor de Vrijheid (PVV) / Chairman Party for Freedom (PVV), Member of Parliament, Netherlands”, which translates to “Chairman parliamentary group Party for Freedom (PVV) / Chairman Party for Freedom (PVV), Member of Parliament, Netherlands.” He is the leader of the extreme right-wing party and is notorious for his firm stance against incoming refugees, both on social media and traditional media.

Lastly, another media voice joins the left-wing force. The account, @volkskrant is a news agency whose stories lobby for human rights. The description reads, “Het officiële Twitter account van de Volkskrant – dagelijks het belangrijkste nieuws en prikkelende opinie. Mail: internet@volkskrantREMOVE_ME.nl. Webcare: @Webcare_VK.” In English this is, “The official Twitter account of the Times - the most important daily news and provocative opinion. Mail: internet@volkskrantREMOVE_ME.nl. Webcare:Webcare_VK.” They are one of the few actors who are on the refugees’ side.

fig. 16: Most Retweeted Users ranked by frequency and categorized ideologically

Even if you go beyond the top ten, one will notice that there will still be more right-wing users than left-wing. This is observable from the table of the 50 Most Retweeted Users seen above (see fig. 16). The column on the far right indicates the user’s political orientation regarding the debate. Blue signifies right-wing, red signifies left-wing, while green indicates that the user is neutral. There is also a distinction made among the media entities (media left-wing, media right-wing, media neutral).

At this point it is clear that within the Actor Composition the dominant voices are extreme right-wing, particularly among the most active and visible users. One can say that in terms of influence the right-wing side is much stronger within the debate. Nine out of the ten most active users are extreme-right wing accounts, and eight of the ten most visible users are of the same orientation. By digging deeper into the profiles of the significant actors, the team determined that the main actors support public protests against refugees and reference right-wing media entities on their tweets or are a media entity itself. A number of these users are also vehemently anti-Islam, with one account claiming to be allergic to Islam, and seem to correlate this religion to terrorism itself. There is also a common sentiment that to be right-wing is to be patriotic to the Netherlands. To conclude this section, the most influential and most active users involved in the debate are extreme right-wing which means that in terms of influence, the right-wing has greater bearing in the refugee issue. They have a louder voice.

Content Analysis

In order to analyse the content, we extracted the Hashtag Frequency, URL Frequency, Media Frequency, and Identical Tweets Frequency. While the previous subsection covered the main actors involved in the refugee issue, this subsection takes a look at the themes, messages, ideas, and the semantic content of the tweets.

Since this research uses Twitter as its main object of study, we can commence the content analysis by studying the main hashtags that were used in relation to the debate. In conducting the hashtag frequency analysis, the researchers noted that one factor to take into account was the short lifespans of these frequently used hashtags. Many of them peak for only one or two days, leading us to question their degree of “popularity”. At the same time, it is not possible to classify all the hashtags according to a stance since most of them are neutral and used by both left and right-wing.

Having said that, the following table displays the hashtags which were most frequently used in the last six months of 2015 (see fig. 17).

fig. 17: Top Hashtags Ranked by Frequency of Usage

It is difficult to say in which context these hashtags are used. Of course #Vluchtelingen is the Dutch word for Refugee, while #AsielZoekers means Asylum Seekers. The hashtags #azc, #AZC, #pvv and #PVV refer to extreme right-wing bodies, while #coa, #COA and #HelpVluchtelingen refer to pro-refugee groups. COA stands for Centraal Orgaan opvang Asielzoekers, or Central Agency for the Reception of Asylum Seekers (“Hervestiging Vluchtelingen”). However, #kominverzet is one of the more important hashtags under study. It is the official resistance hashtag used by right-wing groups. Kom In Verzet is literally a summon to join the resistance. It is a call to action that is being used as a slogan by the right-wing party. It also reflects the effort by the right-wing to frame the conflict as a war, since the need for a resistance presupposes the threat of an invasion or occupation. This goes hand in hand with the right-wing’s appeal to patriotic sentiments, as they build a scenario where their side protects the land from foreigners, in this case refugees.

Steering away from the left-wing versus right-wing dynamic, what is of interest in fig. 17 are the terms that are used to address the issue. Going back to the first research question, to what extent is the refugee issue a crisis, you may notice that #vluchtelingencrisis is more frequently used than #vluchtelingendebat. The data here confirms that three times more people (12,776 vs. 4,159) refer to the issue as a crisis rather than a debate. Therefore we can conclude that Twitter has defined the refugee issue more as a crisis than a mere debate between conflicting ideologies. This gives more gravity to the issue at hand.

There is also the question of how the Twittersphere calls those moving into the country. As migrants, asylum seekers or refugees? From this table we see that the most popular hashtag is #vluchtelingen, or refugees. #Asielzoekers, or asylum seekers, is a far second. #Migranten or migrants is even further down the scale. The fact that more people use the term “refugee” rather than “migrant” is a small win for the left-wing side because the distinction between the two terms comes with serious implications. A migrant is someone who moves to improve their lives by finding work, education or even reuniting with their family (Edwards). On the other hand refugees are “persons fleeing armed conflict or persecution” (ibid.). A refugee is protected by international law, giving them access to asylum procedures and meaning they cannot be sent back to the dangers they have fled (ibid.). Hence the importance of using the correct choice of words.

Moving on to the URL frequency analysis, it is worth noting that 51% of the collection of tweets contain URLs, making it a large portion of the overall data set. Most of these link to relevant media spurring the debate.

fig. 18: Top 50 Media URL’s Ranked by Frequency

All these links lead to images that can be classified according to which political orientation they embody. The following images portray our findings from clicking through the top ten URLs in this list.

fig. 19: Asylum Seekers in the first seven months of 2014 and 2015

The most frequently mentioned URL leads to the table shown above (see fig. 19) that indicates the amount of asylum applications in Europe from 2014 to 2015. This image is used by left-wing actors to highlight how the inflow of refugees and migrants during the first 7 months of 2015 is now at -6% compared to the first 7 months of 2014.

fig. 20: Quote by Rien Vroegindeweij (Dutch poet)

The second most frequently mentioned URL leads to an image that depicts a wall with the quote by Rien Vroegindeweij, “Als iedereen ergens anders vandaan komt is nieman een vreemde.” (see fig. x). This is, “If everyone comes from somewhere else no one is a stranger.” Vroegindeweij is a Dutch poet and writer from Rotterdam, and this wall is located within that city as well, at Kruiskade. The image is circulated by left-wing, pro-refugee users.

fig. 21: Quotes by Loesje

The third and fourth most mentioned URLs, are also material used by the left-wing side (see fig. 21). The images they lead to feature several texts that encourage the welcoming of refugees into the country. These images were created by an organization called Loesje, a group composed of creative writers who embody humanitarian ideals. They write inspiring or provocative texts then self-publish the material online, as well as distribute flyers and posters containing the messages. This group’s materials are taken up by the left-wing voice on Twitter. The fourth URL leads to the image on the right, which reads, “whatever is happening in a country, let children play.”

fig. 22: Media Photo of Refugees in Transit

The fifth URL is a photo of refugees in transit. This image is not particular to the right or left-wing side, as it was used by several new agencies in September.

fig. 23: Photo of Abandoned Hospital in Belgium

The above image also comes up among the top ten URLs. It is an old, abandoned academic hospital in Belgium named Sint Pieter. It is located in the city of Leuven. People on Twitter are suggesting that this could be a suitable venue to house more refugees. We can conclude that the image is used by users who are pro-refugees. However, since the subject is in Belgium the group questions its relevance to the Dutch refugee crisis as it probably only came up because of the use of the Flemish language.

fig. 24: News Article

The seventh in URL frequency ranking is an image of a news article that is used by neither pro nor anti-refugee actors/groups, but by neutral Twitter users. The headline refers to a fight that arose between asylum seekers and inhabitants in the town of Oisterwijk.

fig. 25: “Germany shuts its borders in Austria”

The eighth URL leads to the first material among the top ten URLs that is used by right-wing voices. The text on the image says, “Dramatic development in the Refugee Crisis: Germany shuts its borders in Austria”. This information is used by right-wing Twitter users to strengthen their stance against incoming refugees.

fig. 26: Photos of Female Refugees (left), Female Kurdish MIlitia (right)

The ninth URL leads to the image on the left (see fig. x), which simply depicts female refugees. This is used by left-wing actors. On the other hand, at the bottom of the top ten URLs is an image of female fighters from Iraq who are fighting against ISIS (Webb). They are Kurdish women who form a battalion that is supposedly feared by ISIS. Dying at a woman’s hands forfeits your chance of going to heaven (ibid.). Similar images of these soldiers also show up under frequently mentioned media. The image is used by right-wing voices, possibly in line with their strategy of associating the Middle-East, Islam and refugees with terrorism.

To sum up the URL frequency analysis, six out of the top ten URLs embody the cause of left-wing and pro-refugee users. Two of the URLs are neutral, while only two URLs link to media that push anti-refugee sentiments. Hence, the most frequently mentioned URLs are pro-refugees.

This is very interesting considering that the corpus is against it in general. The right-wing is very active via retweets and thus occupies a vast portion of the Twittersphere. The left-wing is also active even if the total volume of their activity is less than the opposing party. The majority of tweets might even be in favor of refugees, which may explain why the most mentioned URLs are pro-refugee.

This is further ratified by the results of the media frequency analysis. The following table shows the most frequently mentioned media over the last six months of 2015 (see fig. 27).

fig. 27: Media Frequency Table, over time

Many of the URLs that were previously discussed in detail lead to the frequently mentioned media in this section. As is shown above, the largest spikes in the use of media parallel the Twitter activity timeline on page 5, where the peaks occur in September, when Aylan Kurdi’s photo went viral, and mid-October, when Germany and Austria shut their borders. The activity dies down afterwards, and goes slightly up again in December.

Of the most mentioned media in the month of September, two images by Loesje, two of the Belgian hospital, and the Vroegindeweij quote appear on the chart. These are all left-wing and pro-refugee images. The remaining ones are anti-refugee or neutral. In October, two of the three images are left-wing. However, in December 4 more images of the Kurdish female militia circulate Twitter. But overall, the most frequent media are left-wing.

Under content analysis the final aspect of study is the retweet analysis, or identical tweet frequency.

fig. 28: Most Retweeted Tweets

The following figure is a visualization of the top retweets per day (see fig. 29). The list is divided by month and the retweets are resized according to their frequency. The colors are randomly generated to distinguish the tweets. Once again addressing the question “to what extent is the refugee issue a crisis?”, the keywords crisis, debat (debate) and problematiek (problem) were searched among the retweets. Crisis came up the most often, problematiek surfaced once while debat was not. The first user to refer to it as a crisis is the popular radio DJ, Edwin Evers, who encourages his followers to donate money for refugees via the Dutch Red Cross. In mid-October, @geertwilderspvv and another right-wing user, @Gerbijl, use the word in highly retweeted tweets too. Initially we were wondering who benefits the most by calling it a crisis, however this shows that it is termed a crisis by both sides of the conflict.

fig. 29: Top Retweets per Day

Apart from studying how the issue was termed among the top retweets, we were also able to track down the top retweets per month and classify them according to their stance. In July the most retweeted tweet was, RT @GEERTWILDERSPVV: HOPPA! 2000 EXTRA ASIELZOEKERS. VVD = PVDA! #VVDMAAKTNEDERLANDKAPOT HTTPS://T.CO/T4JMWIJSJR. This translates, “Hoppa! 2000 extra asylum seekers. VVD = PVDA! #VVDDestroysNetherlands.” To add to the context, VVD is a slightly right-wing, liberal party while the PVDA is the worker’s party (left-wing).

In August, the most frequently retweeted tweet was: RT @WIMVANDIJCK: STRAKS WORDT ZWARTE PIET NOG DE ENIGE SUKKELAAR DIE NIET MET DE BOOT NAAR EUROPA MAG KOMEN. #ZWARTEPIET #VN #VLUCHTELINGEN. In English, “Soon, Black Pete will be the only low-life who can’t enter Europe by boat #BLACK PETE #UN #REFUGEES”. This is a reference to the anti-racism movement that wanted to ban Black Pete from the Dutch Saint Nicholas feast. The tweet was posted by a right-wing user.

In September, the relevant retweet was: RT @VINCENTMENTZEL: SINDS GISTEREN OP EEN MUUR #KRUISKADE @ROTTERDAM VAN ONZE NEDERLANDSE DICHTER #RIENVROEGINDEWEIJ #VLUCHTELINGEN HTTP://T.CO/8IQ4O1ENQC

This translates, “Since yesterday on a wall #Kruiskade @Rotterdam our Dutch poet #RienVroegindeweij #Refugees”, along with a link leading to the quote on the wall saying “If everyone is coming from somewhere else then no one is a stranger.”

In October, the most frequently retweeted post was, RT @FLIPVANDYKE: WAT ZIJN DE FEITEN AANTALLEN #ASIELZOEKERS? EERSTE ZEVEN MAANDEN 2014 EN 2015 VERGELEKEN. EU BIJNA VERDUBBELD EN NL? HTTP://T.CO/EJIW1VRDSX

In English, “RT @flipvandyke: What are the facts numbers #refugees? First seven months of 2014 and 2015 compared. EU almost doubled, what about the NL? HTTP://T.CO/EJIW1VRDSX”. This was authored by a journalist who is pro-refugee, and he is responsible for one of the most mentioned images during the study.

For November the top retweet was, RT @GEERTWILDERSPVV: BETER: ZE KOMEN ONS LAND NIET IN! WAT ER NIET INKOMT HOEFT ER IMMERS OOK NIET UIT. DUS: NEDERLANDSE GRENZEN DICHT! HTTPS://T.CO/LCKSCISJWL This translates too, “RT @geertwilderspvv: Better: They can’t enter our country. What doesn’t come in, does not have to go out. So: Shut down the Dutch borders”.

Lastly, the top retweet for December was, RT @POLITIE: KOMST #VLUCHTELINGEN LEIDT NIET TOT MEER #CRIMINALITEIT: AFGELOPEN 3 MAANDEN PLEEGDEN ZIJ BEPERKT AANTAL, VOORAL LICHT VERGRIJPEN. This is a tweet from the police, which was quickly picked up by left-wing users. It says, “The influx of #refugees has not led to a spike in crime: in the past 3 months they have committed a limited number of crimes, and the ones they have committed are light offences.”

When it comes to content, it seems that left-wing sentiments are more prevalent. Even during the most significant intervals of Twitter activity, mainly September and mid-October, the dominant retweets and corresponding media content are pro-refugee. This is in stark contrast with the actor composition analysis, where the debate is clearly dominated by right-wing voices. Hence, while the right-wing floods the Twittersphere in terms of influence, the left-wing strongly prevails in terms of content.

Network Analysis

fig. 30: Social Graph by Mention

The nodes on this graph are resized according to how frequently they were mentioned in the data that was gathered. The colors should also distinguish the clusters from one another. Although the labels overlap in many points one can still discern the central nodes, which are @wellofgrief, @azc_alert_db and @azc_alert. The latter two accounts have been given a rich description in the past sections, though the fact that they stand out from the network show how often they have been referenced. This implies that they are the main creators or sources within the discussion.

fig. 31: @Wellofgrief Twitter Profile

On the other hand, the @wellofgrief character has not yet been defined. This account belongs to an extreme right-wing individual who mentions and retweets many right-wing sources like @telegraaf and @azc_alert on his account. He produces angry, sarcastic and provocative tweets against refugees. However, we do not know why more twitter users refer to his account more than official right-wing media accounts. It is possible that for right-wing persons, it is more convenient to visit his account where the anti-refugee content is already concentrated in one place.

All in all, the network graph is quite reflective of the findings in the actor composition analysis, where the most mentioned and retweeted users are extreme-right wing. This visually simplifies the clusters of users that draw and amplify content from the central nodes in the graph above.

Conclusions

By putting together all the findings, one can observe that the most active and visible actors on the network in terms of tweet activity and mentions are extreme right wing voices. Hence, one can say the most influential voices in the debate are extreme right wing and anti-refugees. That is in terms of influence. However, when it comes to overall presence of content, the left-wing has a stronger bearing as seen in the media and URL analyses. The most frequently tweeted and retweeted images and URLs reflect pro-refugee sentiments, especially by the Loesje organization.

These conflicting results can be attributed to the extreme right wing side having more organized and more active user accounts. From the detailed descriptions of each right-wing user in the Actor Composition sub-section, one can pinpoint why exactly they have such a strong influence. Most of the prominent accounts form part of an official body, like @AZC_Alert, @AZC_Alert_DB, and @Telegraaf, which are mass media power houses. At the same time, @geertwilderspvv belongs to a member of parliament who heads an organized movement, the Partij voor de Vrijheid (PVV) or Party for Freedom. In comparison to the individuals and small-scale groups that create and spread left-wing content, it is easy to see how the right-wing voice manages to keep their activity at higher levels.

Reverting back to the main research questions, the first question asks to what extent is the refugee issue a crisis. We found the answer to this question by looking at the top ten hashtags surrounding the debate as well as the top retweets per month. When ranking the hashtags according to their frequency of usage, #vluchtelingencrisis comes out higher than #vluchtelingenproblematiek and #debat. At the same time, among the top retweets, the issue was more commonly referred to as a crisis than as a problematiek. Hence, the issue was elevated to a crisis by the Twittersphere.

The most remarkable discovery from the research actually involves the artificiality of the refugee conversation on Twitter. This is in answer to the second research question, “Is it artificially sustained by an (organised) group of Twitter users of a particular (political) persuasion?” By comparing the Twitter Activity Timeline (fig. 3) with the User Activity graph (fig. x), activity levels generally go down at around the 21st of October. Figure x shows that when this decline happens, certain users like @AZC_Alert_DB continue tweeting in high volumes. Figure 3 also shows that although there is a decline, the activity is still higher than at the beginning, indicating @AZC_Alert_DB’s success in keeping the discussion alive.

The third research question asks “How can one account for the decline of the activity in the Twitter discussion? Does Twitter ‘only’ follow media events?” The team discovered that Twitter activity is spurred by major news events surrounding the debate. These news events include Aylan Kurdi’s death, the EU council’s meeting about the asylum policy, protests in Oranje, Beverwaard and Purmerend, and finally the riots in Geldermalsen, all occurring between July and December 2015. The conversation naturally dies down after the highest peak of activity in mid-October. However, as the twitter activity decreased it was artificially sustained by extreme right wing actors or voices. Therefore we have concluded that Twitter does not only follow media events since certain Twitter users are artificially keeping the debate alive.

Finally, the last research question asks, “Is the decline attributed to the end of the crisis or the end of the organisation of voice? Were particular voices silenced? Was there a conclusion to a debate?”. We can say that the decline in activity may be due to the lack of more news events. The crisis has not come to an end. In truth, the crisis continues but on a lower level of activity.

Recommendations

For researchers who wish to expound on this project, it would be interesting to find a way to apply different measures of importance to the data. These measures could show where else the right-wing and left-wing are leading in the conversation. It can help in better understanding how right-wing actors are more active and visible while left-wing content is actually more present.

The findings have also been very polarized so far. Thus, we recommend a method for measuring or coding the polarization. Within the hashtag analysis, for example, the two camps are either in favor of or against the coming of refugees. Is there a way to measure or code this polarization? If so it would be interesting to measure or extract the non-polarized parts for further analysis. This can also be applied to the polarization of actors within the network.

For a better understanding of the actor composition, it might be worthwhile to look into the @ArmandVervaeck and @PeterAnshof accounts. We have not explained why users are leveraging on these two personalities so much even though they are not official authorities on the issue.

In terms of the hashtag analysis, future researchers could also further explore the relationship between #kominverzet and the right-wing party. We say this because it is one of the more important hashtags, and because it is being used symbolically as the resistance hashtag. There is also a subset of hashtags about Islam that are potentially worth digging into. Although it is a small set, it could help in evaluating if religion is a significant part of the debate.

Bibliography

BBC. “Migrant Crisis: Dutch Town Riots over Asylum Centre Plan.” BBC EuropeBBC News, 17 Dec. 2015. Web. 18 Jan. 2016.

Corder, Mike. “Anger as new migrants sent to tiny Dutch village.” AP. The Big Story, 12 Oct. 2015. Web. 18 Jan. 2016.

Edwards, Adrian. UNHCR viewpoint: “Refugee” or “migrant” - which is right? UNHCR, 27 Aug. 2015. Web. 21 Jan. 2016.

European Council. “Informal meeting of heads of state or government, 23/09/2015.” Consilium Europa. 24 Sept. 2015. Web. 18 Jan. 2016.

“Hervestiging vluchtelingen.” n.d. Web. 22 Jan. 2016.

Loesje. Home Page. Loesje.nl, n.d. Web. 21 Jan. 2016.

Ponsioen, Arnout, and Gijs van Beek. “ ’Over time animation of the evolution of online issue networks".” Digital Methods Initiative - Winterschool 2016. N.p.: Universiteit van Amsterdam, 7 Dec. 2015. Print.

Rogers, Richard. “Otherwise Engaged: Critical Analytics and the New Meanings of Engagement Online.” 11 Jan. 2016. Digital Methods Winter School 2016.

Webb, Sam. “Meet the Fighters Who Strike Fear into the Hearts of ISIS Fanatics.” World news. mirror, 19 Aug. 2015. Web. 21 Jan. 2016.