The Role of Social Network in Information Diffusion

8/22/2019 The Role of Social Network in Information Diffusion

1/26

The Role of Social Networks in

Information Diffusion

Presented by

Xu Zhang


2/26

Paper Information

Authors

Eytan Bakshy (Facebook, Menlo Park, CA, USA ) Social influence and information diffusion in networks

Itamar Rosenn (Facebook, Menlo Park, CA, USA)

Engineering Manager, Measurement Systems at Facebook Cameron Marlow (Facebook, Menlo Park, CA, USA)

Various aspects of online communities

Lada Adamic (University of Michigan, Ann Arbor, MI, USA) Structure and dynamics of social and information networks

Proceedings of the 21st international conference onWorld Wide Web, April 16-20, Lyon, France, 2012.

Times cited: 30


3/26

Content

Motivation

Related work

Experimental design Experimental result

Conclusion

My thought


4/26

Motivation

Social influence can play a crucial role in arange of behavioral phenomena

Dissemination of information

Adoption of political opinions and technologies

Identify the influence remains a challenge

Individuals tend to engage in similar activities as

their peers. Individuals with similar characteristics to associate

with one another.


5/26

Related work

Diffusion and influence have been modeled inblogs, email, and sites (Twitter, Digg, and Flickr)

Correlation between the behavior-engaged friend

number and probability of adopting the behavior Homophily? Peer influence?

Not all diffusion studies must infer whetherone individual influenced another

Do not tell us about the relative importance ofsocial networks in information diffusion


6/26

Experimental design

Experimental platform chose

Design idea

Population Metric


7/26

Experimental design

--Experimental platform chose

A broad online population

Facebook is the most widely used social

networking service in the world. (800million)

Users online personal networks reflect their

real-world connections

Users on average maintain 48% of their real world

contacts on the Facebook.

Interaction among users is well correlated with

self-reported intimacy.


8/26

Experimental design

--Design idea


9/26

Experimental design

--Design idea

Has social networks: OU

(feed mode)

No social networks: U

(no feed mode)

O=OU-U

Information Diffusion

Information Shared

Title: The Role of Social Networks in Information Diffusion


10/26

Experimental design

--Design idea

If a user A shares a URL, will his friend B share

the same URL?

Feed or no feed?

Number of sharing friends?

The relationship between A and B?(Tie strength)


11/26

Experimental design

--Metric

How to judge the influence of social network?

Individual impact: The probability of a user share a

URL

Collective impact: The total incremental quantity

of shared-URLs


12/26

Experimental design

--Population

Random sample (253 million) of all Facebook

users (500 million) who visited the site

between August 14th to October 4th 2010,

and had at least one friend sharing a link.


13/26

Experimental result

Individual impact

Collective impact


14/26

Experimental result

--Individual impact

Feed vs. No feed


15/26

Experimental result

--Individual impact

Number of sharing friends(feed, no feed)


16/26

Number of sharing friends(feed, no feed)

Experimental result

--Individual impact


17/26

Experimental result

--Individual impact

Tie strength (feed, no feed; number shared=1)

2.83

3.84


18/26

Tie strength (feed, no feed; number shared=1)

Experimental result

--Individual impact


19/26

Experimental result

--Collective impact


20/26

Conclusion

Social network can increase the probability

that a user share a URL.

The larger the number of friends that share a

URL, the bigger the probability that a user

share a URL.

Friends that have weak strength have a

significant impact to social network.


21/26

My thought

Information diffusion Information sharing

Social network Facebook

A B C

Content

shared

Content read but

not shared

The Role of Social Networks in Information Diffusion

The Role of Social Networks in Information Sharing within

the Networks


22/26

The End!

Thanks!


23/26


24/26

Experimental design

--Design idea

How to define the tie strength?

Previous work

Number of mutual contacts

This paper () Frequency of private online communication in the form of

Facebook messages

Frequency of public online interaction in the form ofcomments

Number of real-world coincidences in terms of both usersappearing in the same photograph

Number of online coincidences in terms of both usersresponding to the same Facebook post


25/26

Experimental design

--Design idea

How to construct the situation of feed and nofeed?

Feed: normal

No feed: do not expose a user to informationabout their friends sharing behavior

How to ensure the reliability?

Only consider contents shared after the start

Remove URLs that may not reflect ordinary users'purposeful intentions


26/26

: the causal effect of exposure to

information shared by friends with tie

strength , , : the quantity displayed in all users' feeds

posted by friends of tie strength 0 0 ()

=1

Experimental design

--Metric

The Role of Social Network in Information Diffusion

Documents