The Use of Social Media during the 2014 Crisis In Ukraine

THE USE OF SOCIAL MEDIA

DURING THE 2014 CRISIS IN UKRAINE

ANATOLIY GRUZD

@GRUZD

Carolinas ASIST Workshop on Social Media Analysis and Its Application in Research and Practice

October 19, 2015

Associate Professor, Ted Rogers School of Management

Director of the Social Media Lab

Ryerson University

Social Media Lab, Ryerson University (Toronto, Canada)

@gruzd 2

Social Media Support Social Activism and Political Engagement

#OccupyGezi Supporters in Victoria, BC

From this…

Photo credit: Anatoliy Gruzd

In the Social Media Lab,we’re interesting in studying how social media ….

@gruzd 2014 EuroMaidan Revolution

Photo credit: Karl Schönswetter

Social Media Help to Organize Social Movements

… to this

#OccupyGezi: Gezi Parki Protest, Turkey (2013)

4

In the Social Media Lab,we’re interesting in studying how social media ….


Today’s focus:

2014 EuroMaidan Revolution | Revolution of Dignity

5

"2014-02-21 11-04 Euromaidan in Kiev" by Amakuha. Licensed under CC BY-SA 3.0 via Wikimedia

@gruzd

Outline

• Background

• Review of Social Media Platforms

• VK (Vkontakte) Use

• Conclusions

6@gruzd

Today’s focus:

2014 EuroMaidan Revolution | Revolution of Dignity

7

"2014-02-21 11-04 Euromaidan in Kiev" by Amakuha. Licensed under CC BY-SA 3.0 via Wikimedia

November 21, 2013 - Ukraine gov. suspended

the trade & association agreement with EU

@gruzd

February 18-19, 2014: Protests in Kyiv (capital) Turned Deadly

9

source: http://liveuamap.com

RUSSIA

UKRAINE


February 18-19, 2014: Protests in Kyiv Turned Deadly

10

"Barricade line separating interior troops and protesters. Clashes in Kyiv, Ukraine. Events of February 18, 2014-2" by Mstyslav Chernov/Unframe/http://www.unframe.com/ - Licensed under CC

BY-SA 3.0 via Wikimedia Commons


February 18-19, 2014: Protests in Kyiv Turned Deadly

11

"Euromaidan in Kiev 2014-02-19 12-06" by Amakuha. Licensed under CC BY-SA 3.0 via Wikimedia Commons


February-March, 2014: Pro & Anti-Maidan Protests Spread Across

Ukraine

12



February-April 2014: A wave of Anti-Maidan protests in South-East

13@gruzd

Photo credit: Andriy Makukha. Licensed under CC BY-SA 3.0 via Wikimedia Commons

2014 EuroMaidan Revolution

March 2014: Annexation of Crimea by Russia

14



April 2014: Anti-Terrorist Operation (ATO) against pro-Russian Self-proclaimed

republics: Donetsk People’s Republic (DPR) & Lugansk People’s Republic (LPR)

15



Outline

• Background



• Conclusions

16@gruzd 2014 EuroMaidan Revolution

Review of Social Media Platforms:

Websites – still important, linking to social media accounts

http://euromaidansos.org/


http://euromaidansos.org/

Review of Social Media Platforms: Blogs

http://fakecontrol.org/blog/2014/03/04/george-bush/


http://fakecontrol.org/blog/2014/03/04/george-bush/

Review of Social Media Platforms: Blogs Example: Combating Misinformation

http://www.stopfake.org/Source: http://visual.ly/top-7-false-statements-russian-media-about-Ukraine


http://www.stopfake.org/

Review of Social Media Platforms

Youtube & ustream.tv: Activists, News agencies, Gov.officials

https://www.youtube.com/watch?v=I_cNDGU7k98 http://www.ustream.tv/recorded/44404145


Future unrests: Twitter's Periscope?

https://www.youtube.com/watch?v=I_cNDGU7k98

http://www.ustream.tv/recorded/44404145


Instagram: Photo Sharing during Rallies

21

http://www.the-everyday.net


http://www.theguardian.com/world/2014/feb/18/ukraine-police-

storm-kiev-protest-camp-live-updates#start-of-comments


Twitter Widely Used To Disseminate Information about events in Ukraine

Information Diffusion on Twitter



Wikipedia: Editors’ Debate about the Status of Crimea



Facebook Groups & Pages: Activists

https://www.facebook.com/groups/223479324489867 https://www.facebook.com/EvromaidanSOS


https://www.facebook.com/groups/223479324489867

https://www.facebook.com/EvromaidanSOS


Vkontakte & Odnoklassniki.ru (Facebook alike)


Summary: Social Media Use during the 2014 EuroMaidan

Revolution• First “Social Media” Revolution in Ukraine

• 2004 “Orange Revolution” (Pre-social media)

• Internet penetration in Ukraine ~12% (Lysenko & Desouza, 2010)

• Facebook was founded in 2004, Youtube-2005, Twitter-2006, Instagram-2010

• Social media used by both Pro & Anti Maidan activists• Pro-Western vs Pro-Russian groups

• Social media use by governments, elective officials and politicians

• Social media use for “locals” vs. for Westerners (in English)

• Use of multiple social media platforms• Prior social media “revolutions” primarily focused on a single platform: Facebook revolution in Tunisia

(2010/2011); Twitter revolution in Iran (2009/2010)

• Content is often duplicated across multiple platforms


Outline

• Background



• Conclusions


Gruzd, A. & Tsyganova, K. (2014). Politically Polarized Online Groups and

their Social Structures formed around the 2013-2014 crisis in Ukraine.

Internet, Politics, Policy 2014: Crowdsourcing for Politics and Policy. September

25–26, 2014, University of Oxford, Oxford, UK.

This part is based on our recent papers on VK groups

Gruzd, A. & Tsyganova, K. (2015). Information Wars and Online Activism

During the 2013/2014 Crisis in Ukraine: Examining the Social Structures of

Pro- and Anti-Maidan Groups. Policy & Internet 7(2). DOI: 10.1002/poi3.91


http://ipp.oii.ox.ac.uk/2014/programme-2014/track-d-networked-politics/political-and-personal-networks/anatoliy-gruzd-ksenia-tsyganova

http://onlinelibrary.wiley.com/doi/10.1002/poi3.91/abstract

About Vkontakte – #1 Social Networking Website in Ukraine

29

source: http://en.wikipedia.org


Example: VK Group User Interface – Posts, Likes, Comments…


Example: VK Group User Interface – Discussion board, Links & Media Files…


Pro-Maidan (pro-Western) & Anti-Maidan (pro-Russian) Groups on VK

~Over 3,500 groups with membership from 100K+ to under 10 members

32

Anti-Maidan groupPro-Maidan group


Our Research Focus on Online Social Networks

http://www.visualcomplexity.com/vc

Friends’ networks (Facebook,

Twitter, Google+, etc…)

Forum networks

Blog networks

Networks of like-minded people

(YouTube, Flickr, etc…)


1) Represent data as a network

Nodes = People

Edges /Ties = Relations (ex. Who is a friend with whom,

Who replies to whom, etc.)

•2) Apply Social Network Analysis (SNA)

We Study Online Groups from a Network Perspective


• Reduce the large quantity of data into a

more concise representation

• Makes it easier to understand what is

going on in a group

Advantages of Using Social Network Analysis

Once the network is discovered,

we can find out:

• How do people interact with each other,

• Who are the most/least active members of a group,

• Who is influential in a group,

• Who is susceptible to being influenced, etc…


Different Friends’ Networks: What can we learn from them?

36




37



Study Objectives

• Establish baseline measures to interpret emerging social structures of

online activist groups.

• In particular, we ask if there are any observable structural differences or

similarities in social networks formed by VK groups in the two opposing

camps?

• If there are differences, we would like to know whether network properties

alone might be able to suggest a group’s political ideology or other

characteristics.


Data Collection

39

PRO1 PRO2 ANTI1 ANTI2

Num. of Nodes 141,542 96,402 60,506 69,029

Num. of Connections 338,344 221,452 280,678 192,273

• Used VK Public API

• Communities – information about groups and group members

• Wall – posts and comments

• Likes – “likes” that members and visitors leave on posts

• Friends – group members’ friendship relations

• Data collection: 2 most popular (public!) Pro-Maidan and Anti-Maidan groups

• Period: February 18 – May 25, 2015


Method

40

• Social Network Analysis

• SNA measures (e.g., centrality, density, network diameter)

• Exponential Random Graph Modeling (ERGM) – test association tendencies

• Walktrap Community Detection algorithm - identify and describe highly connected

subgroups

• Network Visualization using LGL (Large Graph Layout)

• Manual Content Analysis of

• Group pages and posts

• Sample of public user profiles

• Research software

• Package R (libraries statnet and igraph)

• Tableau for visual analytics


VK Group Example – Pro Maidan #1Friends’ Network (>140k members)

• Formed in early April 2014 to support

Maidan and Antiterrorist Operation

(ATO)

Yellow – users from Ukraine; Red – from Russia; Green – other countries;The layout algorithm is LGL (Large Graph Layout). Isolated nodes are not visible.

0.1

0.2

0.2

0.4

0.4

1.0

1.6

10.2

14.8

69.4

1

2

3

4

5

6

7

8

9

10

0.0 20.0 40.0 60.0 80.0


Subgroup 3

Subgroup 42

VK Group Example – Pro Maidan #1 “Walktrap” Community Detection


Subgroup 3

Subgroup 42

Subgroup 4

Most politically

active

VK Group Example – Pro Maidan #1 “Walktrap” Community Detection

43

Crimean Tatars

Marketing,

Spam



Subgroup 8Subgroup 96



Subgroups 8

Marketing &

spam

Subgroup 96

Marketing & spam


VK Group Example – Pro Maidan #2Friends’ Network

Subgroup 8Subgroup 96

Subgroup 12

Subgroup 3


Subgroup 6

Subgroup 1

VK Group Example –Anti Maidan #1“Walktrap” Community Detection


Subgroup 6

Politically active

users

Subgroup 1:

mixed

“marketing”,

politically

active



Subgroup 6

Subgroup 1

64% from

Donetsk



VK Group Example – Anti-Maidan #2

Friends’ Network (69K members)

One densely connected cluster - suggesting a stronger

agreement among group members from both Ukraine

and Russia

0.1

0.1

0.1

0.3

0.4

0.5

1.1

11.2

34.4

50.0

0.0 10.0 20.0 30.0 40.0 50.0 60.0

1

2

3

4

5

6

7

8

9

10

• In existence since 2011, focused on

Anti-American & Pro-Russia discussions.

• Since the events on Maidan in early

2014, shifted its focus to support Anti-

Maidan activism & the two self-

proclaimed republics – Donetsk and

Lugansk People's Republics.







Politically active

accounts

Politically active


Conclusions - Comparing Groups across Political Divide


Number of Nodes 141,542 96,402 60,506 69,029

Number of Connections 338,344 221,452 280,678 192,273

Network Diameter 19 19 16 21

Clustering Coefficient 0.08 0.09 0.09 0.13

Degree Centralization 0.0048 0.0073 0.0815 0.0260

Modularity Index 0.58 0.58 0.42 0.24

%Users from Ukraine 69 73 57 34

%Users from Russia 15 10 30 50


Conclusions - Geography Matters!

• Although all four groups included

people from both Ukraine and

Russia, the ERGM models

confirmed the tendency of group

members to friend others in the

same country.

• Furthermore, we also observed

homophily among users from

the same city for the top-10 cities

with the most number of VK users

in all groups

54

Online social networks likely represent local and potentially

pre-existing social networks

"Euromaidan Protests" by Lvivske, NickK - Sources for particular cities are given at w:uk:Євромайдан

у регіонах України. Licensed under CC BY-SA 3.0 via Wikimedia Commons


Conclusions - Comparing Groups across Political Divide


Number of Nodes 141,542 96,402 60,506 69,029

Number of Connections 338,344 221,452 280,678 192,273

Network Diameter 19 19 16 21

Clustering Coefficient 0.08 0.09 0.09 0.13

Degree Centralization 0.0048 0.0073 0.0815 0.0260

Modularity Index 0.58 0.58 0.42 0.24

%Users from Ukraine 69 73 57 34

%Users from Russia 15 10 30 50


Conclusions - Position of Spam & Marketing Accounts

• In the PRO1 and PRO2 groups, spam & marketing accounts appeared to be organized in a densely connected subgroup with high degree centrality values and low user engagement

• It is important to differentiate spam & marketing accounts from group supporters who might also be located in smaller, somewhat isolated subgroups because of their minority status in the group (e.g., Crimean Tatars in PRO1)

• Interestingly, this pattern was not observed in the ANTI1 or ANTI2 groups.



57

Pro-Maidan Groups Anti-Maidan Groups

Study Objectives (recap):

• Establish baseline measures to interpret emerging social structures of online activist groups.

• In particular, we ask if there are any observable structural differences or similarities in social networks formed by VK groups in the two opposing camps?

• If there are differences, we would like to know whether network properties alone might be able to suggest a group’s political ideology or other characteristics.



58

Pro-Maidan Groups Anti-Maidan Groups

Conclusions:

• Group-level indicators, such as degree centralization, modularity index and average engagement level, could help to classify and describe groups based on their network properties and structures.

• BUT: SNA alone could not explain the structural differences between the four networks we examined.

• SOLUTION: a combination of SNA, visualization and community detection algorithm, coupled with a manual content analysis of a sample of group messages and user profiles is an effective approach to study the underlying social structures of online activist groups.


Photo credit: Karl Schönswetter

#OccupyGezi: Gezi Parki Protest, Turkey (2013)

59

#OccupyGezi Supporters in Victoria, BC

?

Study Implications: Can we predict successful groups? Can we predict group’s longevity?


Future Work

• Explore the roles of isolates in the

group

– lurkers, spammers, trolls, or fake

accounts?

60

?

• Investigate how people from

countries other than Ukraine

and Russia participated in these

groups


Acknowledgments

• We thank Emad Khazraee, Dmitri Tsyganov, Andrea Kampen,

Philip Mai, and Elizabeth Dubois for their help in preparing this

article.


THE USE OF SOCIAL MEDIA

DURING THE 2014 CRISIS IN UKRAINE

ANATOLIY GRUZD

@GRUZD

Carolinas ASIST Workshop on Social Media Analysis and Its Application in Research and Practice

October 19, 2015

Associate Professor, Ted Rogers School of Management

Director of the Social Media Lab

Ryerson University