+ All Categories
Home > Documents > Using Geo-Located Social Media Data to Study Refugee Crises · 2017. 5. 24. · Using Geo-Located...

Using Geo-Located Social Media Data to Study Refugee Crises · 2017. 5. 24. · Using Geo-Located...

Date post: 15-Oct-2020
Category:
Upload: others
View: 0 times
Download: 0 times
Share this document with a friend
3
Using Geo-Located Social Media Data to Study Refugee Crises Jamie Mahoney, Shaun Lawson, Tom Feltwell Northumbria University Newcastle upon Tyne, UK [email protected] Christoph Scheib University of Kaiserslautern Kaiserslautern, Germany [email protected] ABSTRACT Previous research has utilized social media data as a means to observe, and gain insights into, particular populations, or population groups including those subject to marginalisation and stigmatisation. In this paper, we discuss our exploratory research that investigates the use of visual social media as a mechanism for documenting the experiences of refugees moving through Europe and the attitudes of European citizens from different nations towards the refugees. We briefly outline our initial motivations for this work, the methods used, the problems experienced and the implications for potential future research in this area. KEYWORDS Refugees, social media, Instagram, visual analysis. 1 INTRODUCTION There is a good deal of recent research that has utilized social media platforms and social media data as a resource for observing, and gaining insights into, populations or population groups within specific contexts [e.g. 2, 3, 4, 8, 9]. This includes work that has examined representations of those subject to marginalisation and stigmatisation through the media [2, 3] and political representation [4, 9]. Building on the approaches taken in this previous work, we investigated the use of Instagram, primarily a visual-focused social media platform, as a means of exploring, documenting and sharing the experiences of, and attitudes towards, refugees and their movement through Europe. With platforms such as Instagram providing geo-location data attached to many posts 1 , our initial research questions were centered on the premise that we may be able to find examples of refugees sharing images online, and associate these images with both particular geographical regions and distinct time frames. This assumption was based on the well documented high penetration rate of smartphone technology across the world [11]. However, our initial assumptions proved, in hindsight, to be extraordinarily naïve; what we actually found was that the vast majority of geo-located posts, broadly following the refugee ‘corridor’ through Europe (as shown in Figure 1), were posted by a combination of Instagram users who were either (i) European citizens who had not been displaced, or (ii) members of the media, including photojournalists. We therefore sought to modify our research questions and present our current thinking here for further dissemination and discussion. 1 depending on a user’s location sharing settings In the remainder of this paper, we therefore outline our approach to data collection and our initial analyses. We then conclude with a discussion of our experiences of collecting and analyzing this kind of social media data, and the considerations for future work in this and related areas. 2 DATA COLLECTION Using Instagram’s ‘Tags’ API endpoint [7], we collected all public posts that were primarily shared with the hashtag #refugee, as well as associated and derivate hashtags such as #refugees, dating between March 2011 and March 2016. The entire collected dataset consisted of over 150,000 images, as well as their associated captions, posted in a range of different languages, meta-data and comments; approximately 30,000 of these images had geo-location data associated with them. As a case study, we focused on geo- located posts within Europe, as since 2011 the civil war in Syria has caused the displacement of approximately 10 million people, with an estimated 4.5 million Syrian refugees fleeing to neighbouring countries [1]. The heatmap in Figure 1 shows the geographical areas in Europe that demonstrated high posting activity; these areas broadly map to the migration ‘corridor’ as detailed by the UNHCR [10]. Whilst there are many locations in which Syrian refugees have sought asylum, since 2014 a prominent destination was Germany, and the preferred route as through Turkey, across the Balkan countries and northwards into Germany. This route crosses multiple national borders, and in response to the Figure 1. Instagram activity reflecting refugee movement through Europe.
Transcript
Page 1: Using Geo-Located Social Media Data to Study Refugee Crises · 2017. 5. 24. · Using Geo-Located Social Media Data to Study Refugee Crises Jamie Mahoney, Shaun Lawson, Tom Feltwell

Using Geo-Located Social Media Data to Study Refugee Crises

Jamie Mahoney, Shaun Lawson, Tom Feltwell Northumbria University

Newcastle upon Tyne, UK [email protected]

Christoph Scheib University of Kaiserslautern

Kaiserslautern, Germany [email protected]

ABSTRACT Previous research has utilized social media data as a means to observe, and gain insights into, particular populations, or population groups including those subject to marginalisation and stigmatisation. In this paper, we discuss our exploratory research that investigates the use of visual social media as a mechanism for documenting the experiences of refugees moving through Europe and the attitudes of European citizens from different nations towards the refugees. We briefly outline our initial motivations for this work, the methods used, the problems experienced and the implications for potential future research in this area.

KEYWORDS Refugees, social media, Instagram, visual analysis.

1 INTRODUCTION There is a good deal of recent research that has utilized social media platforms and social media data as a resource for observing, and gaining insights into, populations or population groups within specific contexts [e.g. 2, 3, 4, 8, 9]. This includes work that has examined representations of those subject to marginalisation and stigmatisation through the media [2, 3] and political representation [4, 9]. Building on the approaches taken in this previous work, we investigated the use of Instagram, primarily a visual-focused social media platform, as a means of exploring, documenting and sharing the experiences of, and attitudes towards, refugees and their movement through Europe. With platforms such as Instagram providing geo-location data attached to many posts 1 , our initial research questions were centered on the premise that we may be able to find examples of refugees sharing images online, and associate these images with both particular geographical regions and distinct time frames. This assumption was based on the well documented high penetration rate of smartphone technology across the world [11]. However, our initial assumptions proved, in hindsight, to be extraordinarily naïve; what we actually found was that the vast majority of geo-located posts, broadly following the refugee ‘corridor’ through Europe (as shown in Figure 1), were posted by a combination of Instagram users who were either (i) European citizens who had not been displaced, or (ii) members of the media, including photojournalists. We therefore sought to modify our research questions and present our current thinking here for further dissemination and discussion. 1 depending on a user’s location sharing settings

In the remainder of this paper, we therefore outline our approach to data collection and our initial analyses. We then conclude with a discussion of our experiences of collecting and analyzing this kind of social media data, and the considerations for future work in this and related areas.

2 DATA COLLECTION Using Instagram’s ‘Tags’ API endpoint [7], we collected all public posts that were primarily shared with the hashtag #refugee, as well as associated and derivate hashtags such as #refugees, dating between March 2011 and March 2016. The entire collected dataset consisted of over 150,000 images, as well as their associated captions, posted in a range of different languages, meta-data and comments; approximately 30,000 of these images had geo-location data associated with them. As a case study, we focused on geo-located posts within Europe, as since 2011 the civil war in Syria has caused the displacement of approximately 10 million people, with an estimated 4.5 million Syrian refugees fleeing to neighbouring countries [1]. The heatmap in Figure 1 shows the geographical areas in Europe that demonstrated high posting activity; these areas broadly map to the migration ‘corridor’ as detailed by the UNHCR [10]. Whilst there are many locations in which Syrian refugees have sought asylum, since 2014 a prominent destination was Germany, and the preferred route as through Turkey, across the Balkan countries and northwards into Germany. This route crosses multiple national borders, and in response to the

Figure 1. Instagram activity reflecting refugee movement through Europe.

Page 2: Using Geo-Located Social Media Data to Study Refugee Crises · 2017. 5. 24. · Using Geo-Located Social Media Data to Study Refugee Crises Jamie Mahoney, Shaun Lawson, Tom Feltwell

2

movement of thousands of migrants, many countries along the route put measures in place, at various times, to stop the flow of refugees and close the borders. Ultimately, in March 2016, Slovenia closed their border to refugees, and thus the Western Balkans route was deemed closed [6], hence we did not collect data after this point.

3 DATA ANALYSIS Casual manual browsing of the collected dataset revealed a rich set of visual imagery reflecting the complexity of the real world issues associated with refugee movement through Europe. We initially therefore struggled with the sheer size of the dataset; with the entire dataset covering a 5-year period from 2011 to 2016, we first segmented the dataset into subsets based on the time the posts were created, with each subset covering a 7-day period. Visualizing the locations of posts sent within each 7-day period allowed us to see how the posting patterns developed over time. Figure 2 shows a heatmap for two different timeframes, with the location of posts moving up through Europe as time progressed. By viewing the changes in posting locations over time, the data seemed to reinforce our initial thinking that social media use was ‘following’ refugees as they moved throughout Europe. Additionally, major events that would affect refugees’ movements also coincided with increased posting activity in the affected region. To further reinforce our original thoughts, tracing individuals’ posting behaviors suggested that postings were “moving through Europe” and occurring in different locations as time progressed. We then however began to explore, in detail, the content of the images themselves. We quickly realized that without further investigation of the accounts that were posting, it would have been very easy to make assumptions that would have led to erroneous findings. This highlights the need to avoid treating social media data collection and analysis as a ‘quick and easy’ approach: while data collection may be simple, understanding the individuals behind the social media accounts can be far more complex. For example, in our dataset there were examples of posts sent from the general public in regions directly affected by the movement of refugees (Figure 3 (i)), and those created by professional photojournalists (Figure 3 (ii)). While we initially assumed to see some evidence of refugees documenting their experiences, all of

these images were created and shared by other parties – be it local populations, or third-parties following the refugee crisis, such as the media. The absence of refugees as users within the dataset is may be attributed to the fear of surveillance. Gillespie [5] notes how public social media and GPS traces leave information that could be used to surveil or track them, and as such refugees, whilst embracing the empowering technologies of the smartphone, are wary of public sharing via smartphones, and thus prefer encrypted communications.

4 DISCUSSION This paper describes our initial, and somewhat naïve, approach to exploring a rich visual dataset gathered during a large scale refugee movement; whilst initial hypotheses would indicate the presence of data produced by refugees, our analysis demonstrates a more nuanced constitution, primarily composed of those observing the refugee movements, be they members of the public in locations moved through by refugees, or journalists purposefully following their movements. As a result of our initial findings in this area, we highlight aspects that should be given very careful consideration when implementing social media based research with specific populations, such as refugees. A point which applies to a far wider range of research is the need to avoid taking social media data at ‘face value’. Given our initial assumptions, it would have been very easy to believe that the collected data supported our initial hypothesis, with posting trends moving through Europe as time progressed, and individual users posting at several points throughout the refugee pathways. It was only in taking a holistic view of the data, including metadata such as user account details, that our initial hypothesis was challenged, and prompted deeper reflection and analysis. Geographically distributed data of this type could be used to explore the rich differences in empathy across multiple countries and time periods, as well as allowing the comparison of overall sentiment towards and about refugees.

The need to treat the abundance of data made available through social media platforms with caution is a further challenge in this area. As demonstrated in the study briefly outlined here, it is straightforward to collect a significant dataset, including data sent from thousands of accounts, impacting a large population of

Figure 2. Two heatmaps showing posting locations, separated by 7 days.

Figure 3. Examples of posts shared by (i) general public and

(ii) photojournalists

(i) (ii)

Page 3: Using Geo-Located Social Media Data to Study Refugee Crises · 2017. 5. 24. · Using Geo-Located Social Media Data to Study Refugee Crises Jamie Mahoney, Shaun Lawson, Tom Feltwell

3

people. Whilst a cursory analysis of that data could be used to support (or contradict) initial hypotheses, developing an understanding of the individuals ‘behind’ the accounts doing the posting is a far more complex process. The density of visual media, the associated metadata with each account, combined with the large quantity of data that can be obtained from social media platforms presents methodological challenges, which are being actively explored [9]. We intend to extend this work with a full study encompassing the geographical and temporal aspects of the data, which may reveal differences in empathy, distrust and marginalisation.

REFERENCES [1] BBC News. Syria – The story of the conflict.

http://www.bbc.co.uk/news/world-middle-east-26116868 [2] Phil Brooker, John Vines, Selina Sutton, Julie Barnett, Tom

Feltwell, and Shaun Lawson. 2015. Debating Poverty Porn on Twitter: Social Media as a Place for Everyday Socio-Political Talk. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI '15). ACM, New York, NY, USA, 3177-3186. DOI: https://doi.org/10.1145/2702123.2702291

[3] Mark Doughty, Shaun Lawson, Conor Linehan, Duncan Rowland, and Lucy Bennett. 2014. Disinhibited abuse of othered communities by second-screening audiences. In Proceedings of the ACM International Conference on Interactive Experiences for TV and Online Video (TVX '14). ACM, New York, NY, USA, 55-62. DOI: https://doi.org/10.1145/2602299.2602311

[4] Tom Feltwell, Jamie Mahoney, and Shaun Lawson. 2015. "Aye, have a dream #IndyRef": use of instagram during the scottish referendum. In Proceedings of the 2015 British HCI Conference (British HCI '15). ACM, New York, NY, USA, 267-268. DOI: http://dx.doi.org/10.1145/2783446.2783604

[5] Mary Gillespie. Phones – crucial to survival for refugees on the perilous route to Europe.2016. The Conversation. Accessed 10th May 2017 from http://theconversation.com/phones-crucial-to-survival-for-refugees-on-the-perilous-route-to-europe-59428

[6] Guardian. 2016. Balkan Refugee Route Closed https://www.theguardian.com/world/2016/mar/09/balkans-refugee-route-closed-say-european-leaders

[7] Instagram. 2017. API Endpoints. https://www.instagram.com/developer/endpoints/

[8] Gilad Lotan et al. "The Arab Spring| the revolutions were tweeted: Information flows during the 2011 Tunisian and Egyptian revolutions." International journal of communication 5 (2011): 31.

[9] Jamie Mahoney, Tom Feltwell, Obinna Ajuruchi, and Shaun Lawson. 2016. Constructing the Visual Online Political Self: An Analysis of Instagram Use by the Scottish Electorate. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems (CHI '16). ACM, New York, NY, USA, 3339-3351. DOI: https://doi.org/10.1145/2858036.2858160

[10] United Nations High Commission for Refugees. 2017. Syrian Regional Refugee Response. http://data.unhcr.org/syrianrefugees/regional.php

[11] World Bank. 2017. Mobile Cellular Subscriptions per 100 People.http://data.worldbank.org/indicator/IT.CEL.SETS.P2


Recommended