1

Finding Political Network Bridges on FacebookNasri Messarra, Anne MioneUniversité de Montpellier 1 - MRM

2 Background

When two separate clusters possess non-redundant information, there is said to be a structural hole between them (Burt, 1992)

A network that bridges structural holes will provide network benefits that are in some degree additive, rather than overlapping (Burt, 2004)

“Brokerage across the structural holes between groups provides a vision of options otherwise unseen, which is the mechanism by which brokerage becomes social capital” (Burt, 2004)

“The between-group brokers are more likely to express ideas, less likely to have ideas dismissed, and more likely to have ideas evaluated as valuable.” (Burt, 2004).

Advantages of bridging Structural Holes

3Structural holes and bridges in the context of polarized crowds

Polarized Crowd (Smith, 2014): Two big and dense groups Little or no connection between

them despite the fact that they are focused on the same topic

The topics being discussed are often highly divisive and heated political subjects.

Ignoring one another

How to identify bridges in the context of polarized crowds?

Question

4 Research Question

Is it possible to use Facebook to identify bridges overlapping structural holes in polarized crowds? Experimenting on a political situation

5 Literature Review:Politics and OSN Political science is fascinated with networks (Ward & al.,

2011) Online social networks are now an essential part of any

political campaign (Mercanti-Guerin, 2010) Online social networks are used as a news or rumor diffusion

tool through WoM (Berinsky, 2012; Finke & Harris, 2012; Huang, 2013)

On OSN politicians use a “politics of threat” to change Citizens’ perception regarding foreign policy views. Emotional status influence views even when the facts are the same (Gadarian, 2010)

Emergence and creation of centrally-controlled accounts to create the appearance of widespread support for a candidate or opinion (Ratkiewicz & al., 2011)

6Polarized Crowds and Structural holes: Lebanon’s situation

March 8 March 14

- Lebanese division between the two major political movements, March 8 and March 14, has reached a dramatic point where all the political and economical activity is literally paralyzed

- Media (newspapers, televisions, radio) are owned by political figures and only give their own interpretation of every event, increasing the gap between the two movements.

- A few movements strive to initiate a change in directions but they represent a very small minority of the Lebanese population.

7Polarized crowd: The Lebanon’s case

An experiment on Twitter with more than 3,000 tweets (in Arabic, French, English and Turkish) showed that, even regarding two major events on which both movements strongly disagreed: The trial of Michel Samaha, an

ex-minister (March 8) accused of terrorist acts,

The involvement of Hezbollah (March 8) in the Qalamun battle and its repercussions,

both movements did not engage with each others

8

Methodology• We use Facebook because it is the largest OSN and a network of “friends”• We use Facebook timelines (not pages) because timelines provide access to friends

relations while Facebook pages and Facebook groups have limited functionality.• We use the network a third profile, a politician that is not aligned and is himself a

bridge between the two major political movements to increase the chances of finding friends in his networks that bridge over structural holes.

• We use the profiles of:• March 14 politician with 994 Facebook friends • March 8 politician with 4,889 friends• The profiles of “M”, a non aligned politician (8,440 unique friends).

• Using OpenGraph, NodeXL and Gephi, we draw the edges (friends of friends) and combine the 3 networks to find Facebook users who belong to the M.A. group but have friends in the other 2 groups (March 8 and March 14)

March 14

March 8

M

9Methodology: Processing Data

Start

Select a friend of M.

F1>0

Count his friends in the March 8 group

(F1)

Count his friends in the March 14 group

(F2)

F2>0

Yes

Yes

Score = Score + 10

Score = 100

No

Score = Score + 1

Save Score, F1, F2

F1>F2

No

OrangeColor

BlueColor

Last Friend?

End

Yes

No

1. Retrieve all friends from the politicians profiles using NodeXL

2. Scrap all friend to friend connections

3. Compile all profiles into one workbook

4. Calculate a “bridge value” (or score) based on the groups were the user is in or has friends in (vertex 1 or vertex 2):• M = 100; M8 = 10; M14 = 1• M and M8 = 101• M and M14 = 110• M8 and M14 = 11• M, M8 and M14 = 111

10

• From the compiled population of the 3 politicians (15,344 unique vertices and 341,345 edges),

• We find 807 unique bridges as follows:

Results: 807 bridges

  M M8 M14 Count

Bridge Between M and M8

1 1 0 622

Bridge Between M and M14

1 0 1 99

Bridge Between M8 and M14

0 1 1 52

Bridge Between M, M14 and M8

1 1 1 34

Total 807

11Results: Filtering data and investigating manually the most connected bridges• We filter the bridges,

removing all non-bridges

• The bridges are themselves connected as friends (graph density=4.15%)

• The bridges with more than 100 connections are well known journalists, judges, businessmen, lawyers, etc. Most of them with no declared affiliation to one of the two main political movements in Lebanon.

Network graph of political bridges with more than 100 connections, color-coded.

12 Contribution to Viral Marketing Hinz, Skiera, Barrot & Becker (2011) define 4 critical viral marketing success

factors: 1. Content, in that the attractiveness of a message makes it memorable

(Berger and Milman, 2011; Berger and Schwartz, 2011; Gladwell, 2002; Porter and Golan, 2006)

2. The structure of the social network (Bampo et al. 2008) 3. The behavioral characteristics of the recipients and their incentives for

sharing the message (Ardnt, 1967); 4. The seeding strategy which determines the initial set of targeted

consumers chosen by the initator of the viral campaign (Bampo et al., 2008; Kalish, Mahajan, and Muler, 1995; Libai, Muller and Peres, 2005). Several researchers agree that the size of the initial seeding population is not important, on the contrary. The marketer should focus on targetting a small population of influentials or easily influenced people (Scarpi, 2010; Wallace, Buil, Chernatony, 2014; Watts & Dodds, 2007; Liu-Thompkins, 2012)

13Contributions to Political Marketing and Structural Holes

We looked at political marketing on OSN from a new and broader perspective and analyzed the electorate on OSN not as separate groups but as a connected social graph of supporters, opponents and political bridges.

While the focus has been on the message as news or rumors (Berinsky, 2012; Finke & Harris, 2012; Huang, 2013) and the entertaining “wrap” around it (Carpini & Williams, 2001), or the timing and the psychology of the recipients (Gadarian, 2010), we focus on the role of the recipients and their position.

We were able to show that it is possible for a marketer to detect bridges connecting one or more networks and to eventually understand the information flow, not only in his network but in the larger population.

OSN can be used as a tool to determine scientifically between-group brokers (bridges) and interpersonal ties.

 

14 Conclusion Our method could be used as a first step to define an initial seeding

population (Bampo et al., 2008; Kalish, Mahajan, and Muler, 1995; Libai, Muller and Peres, 2005) and start a viral marketing campaign.

With the larger perspective offered by a combined graph of several groups of politically engaged people, the marketer could choose to target hubs to diffuse information to his supporters quickly or target bridges to influence or diffuse information to the opposition.

Having identified bridges, we can anticipate some of their behavioral characteristics (Ardnt, 1967; Hinz, Skiera, Barrot & Becker, 2011) and monitor them to see how they react (or not) to messages.

15 Discussion This experiment helped locate political bridges between the different

political networks on Facebook. Yet, we don’t know if these bridges will be willing to diffuse political information, create a snowball effect and if others are or will be influenced by them.

It would be interesting to push this experiment further and understand what type of information these bridges share on Facebook and Twitter and understand their behavior.

While this experiment can be reproduced today, Facebook is continuously limiting one’s access to his own data (fans, friends of friends, etc.) and this experiment may not be possible in the future.

16 Thank You

Nasri Messarra: nasri@messarra.com

Anne Mione: anne.mione@univ-montp1.fr