Social Media Mining: An Introduction

P1: qVa Trim: 6.125in×9.25in Top: 0.5in Gutter: 0.75in
CUUS2079-07 CUUS2079-Zafarani 978 1 107 01885 3 January 13, 2014 17:17

7.5 Summary 209

100

80

60

Population 40

20

0

0102030

t

40 50

S(t)

I(t)

R(t)

Figure 7.17. SIRS Model Simulated withS 0 =99,I 0 =1,R 0 =0,γ= 0 .1,β= 0 .01,

andλ= 0 .02.

entails vaccinating a population inside a herd such that the pathogen cannot

initiate an outbreak inside the herd. In general, creating herd immunity

requires at least arandomsample of 96% of the population to be vacci-

nated. Interestingly, we can achieve the same herd immunity by making use

of friends in a network. In general, people know which of their friends have

more friends. So, they know or have access to these higher degree and more

connected nodes. Christakis found that if a random population of 30% of

the herd is selected and then these 30% are asked for their highest degree

friends, one can achieve herd immunity by vaccinating these friends. Of

course, older intervention techniques such as separating those infected from

those susceptible (quarantining them) or removing those infected (killing

cows with mad cow disease) still work.

7.5 Summary

In this chapter, we discussed the concept of information diffusion in social

networks. In the herd behavior, individuals observe the behaviors of others

and act similarly to them based on their own benefit. We reviewed the

well-known diners example and urn experiment and demonstrated how

conditional probabilities can be used to determine why herding takes place.

We discussed how herding experiments should be designed and ways to

intervene with it.