rumour spreading in social networks

77
Rumour Spreading in Social Networks Alessandro Panconesi Dipartimento di Informatica Joint work with Flavio Chierichetti and Silvio Lattanzi

Upload: wesley

Post on 09-Jan-2016

59 views

Category:

Documents


1 download

DESCRIPTION

Rumour Spreading in Social Networks. Alessandro Panconesi Dipartimento di Informatica Joint work with Flavio Chierichetti and Silvio Lattanzi. Rumours spread quickly. OUR GOAL. Argue in a rigorous way that rumours spread quickly in a social network. How to tackle the problem. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Rumour Spreading in Social Networks

Rumour Spreading in Social Networks

Alessandro PanconesiDipartimento di Informatica

Joint work with Flavio Chierichetti and Silvio Lattanzi

Page 2: Rumour Spreading in Social Networks

Rumours spread quickly

Page 3: Rumour Spreading in Social Networks

OUR GOAL

Argue in a rigorous way that rumours spread

quickly in a social network

Page 4: Rumour Spreading in Social Networks
Page 5: Rumour Spreading in Social Networks

How to tackle the problem

Page 6: Rumour Spreading in Social Networks

How to tackle the problem

Page 7: Rumour Spreading in Social Networks

OUR GOAL

Prove that rumours spread quickly in a

social network

Page 8: Rumour Spreading in Social Networks

Gossip: a very simple model

Page 9: Rumour Spreading in Social Networks

Gossiping

Page 10: Rumour Spreading in Social Networks

Gossiping

Page 11: Rumour Spreading in Social Networks

Gossiping

Page 12: Rumour Spreading in Social Networks

Gossiping

Page 13: Rumour Spreading in Social Networks

Gossiping

Page 14: Rumour Spreading in Social Networks

Gossiping

Page 15: Rumour Spreading in Social Networks

Gossiping

Page 16: Rumour Spreading in Social Networks

Gossiping Variants

PUSH Node with information sends toa random neighbour

Page 17: Rumour Spreading in Social Networks

Gossiping Variants

PUSH

PULL

Node with information sends toa random neighbour

Node without information asks a random neighbour

Page 18: Rumour Spreading in Social Networks

Gossiping Variants

PUSH

PULL

Node with information sends toa random neighbour

Node without information asks a random neighbour

PUSH-PULL

Page 19: Rumour Spreading in Social Networks

Motivation• Technological: Rumour spreading

algorithms are widely used in communication networks which, more and more, are likely to exhibit a social dimension. This knowledge might be exploited for more efficient communication protocols

• Sociological: rumour spreading is a basic, simple form of a contagion dynamics. By studying it we hope to gain some insight into more complex diffusion phenomena

Page 20: Rumour Spreading in Social Networks

Previous Work

Page 21: Rumour Spreading in Social Networks

Different approach

We are looking for necessary and/or sufficient conditions for rumour spreading to be fast in a given network

Page 22: Rumour Spreading in Social Networks

Push

Page 23: Rumour Spreading in Social Networks

Push

Page 24: Rumour Spreading in Social Networks

Push

Page 25: Rumour Spreading in Social Networks

Push

Page 26: Rumour Spreading in Social Networks

Push

Page 27: Rumour Spreading in Social Networks

Push

Page 28: Rumour Spreading in Social Networks

Pull

Page 29: Rumour Spreading in Social Networks

Pull

Page 30: Rumour Spreading in Social Networks

Pull

Page 31: Rumour Spreading in Social Networks

Pull

Page 32: Rumour Spreading in Social Networks

Pull

Page 33: Rumour Spreading in Social Networks

Pull

Page 34: Rumour Spreading in Social Networks

Both Push and Pull are hopeless

Page 35: Rumour Spreading in Social Networks

Therefore, we consider Push-Pull, quite

appropriately in the Age of the Internet

Page 36: Rumour Spreading in Social Networks

Push

Page 37: Rumour Spreading in Social Networks

Push Pull

Page 38: Rumour Spreading in Social Networks

Push Pull

Push-Pull

Page 39: Rumour Spreading in Social Networks

OUR GOAL

Prove that rumours spread quickly in a

social network

Page 40: Rumour Spreading in Social Networks

Time is of the essence

Page 41: Rumour Spreading in Social Networks

Gossiping

0

Page 42: Rumour Spreading in Social Networks

Gossiping

1

Page 43: Rumour Spreading in Social Networks

Gossiping

1

Page 44: Rumour Spreading in Social Networks

Gossiping

1

Page 45: Rumour Spreading in Social Networks

Gossiping

2

Page 46: Rumour Spreading in Social Networks

Gossiping

2

Page 47: Rumour Spreading in Social Networks

Gossiping

3

Page 48: Rumour Spreading in Social Networks

Time is of the essence

Time = #rounds

Speed = Time ispoly-logarithmic

Page 49: Rumour Spreading in Social Networks

OUR GOAL

Prove that rumours spread quickly in a

social network

Page 50: Rumour Spreading in Social Networks

Problem formulation:

How many rounds will it take Push-Pull to broadcasta message in a social network?

Prove that rumours spread quickly in a

social network

Recall our goal..

Page 51: Rumour Spreading in Social Networks

But..what is a social network??

Page 52: Rumour Spreading in Social Networks

Argue about a model

Page 53: Rumour Spreading in Social Networks

Argue about a model

• Chierichetti, Lattanzi, P [ICALP’09] Randomized broadcast is fast in PA graphs: with high probability, regardless of the source, push-pull broadcasts the message within O(log2N) many rounds

Page 54: Rumour Spreading in Social Networks

Argue about a model

• Chierichetti, Lattanzi, P [ICALP’09] Randomized broadcast is fast in PA graphs: with high probability, regardless of the source, push-pull broadcasts the message within O(log2N) many rounds

• Dörr, Fouz, Sauerwald [STOC’11] show optimal Θ(logN) bound holds

Page 55: Rumour Spreading in Social Networks

Argue about a model

However, there is no accepted model for social networks…

Page 56: Rumour Spreading in Social Networks

Empiricism to the rescue

Leskovec et al [WWW’08] show that real-world networks (seem to) enjoy high conductance (in the order of log -1 N)

Page 57: Rumour Spreading in Social Networks

ConductanceS

Page 58: Rumour Spreading in Social Networks

ConductanceS

Page 59: Rumour Spreading in Social Networks

ConductanceS

Page 60: Rumour Spreading in Social Networks

ConductanceS

Page 61: Rumour Spreading in Social Networks

ConductanceS

Page 62: Rumour Spreading in Social Networks

Our Goal finally becomes..

Prove that if a network has high conductance then rumours spread

quickly

Page 63: Rumour Spreading in Social Networks

Our Goal finally becomes..

Prove that if a network has high conductance then rumours spread quickly assuming a worst case source

Page 64: Rumour Spreading in Social Networks

ResultsChierichetti, Lattanzi, P [SODA’10] With high

probability, regardless of the source, push-pull broadcasts the message within

O(log4 N/ 6) many rounds

Chierichetti, Lattanzi, P [STOC’10] Improved to O( -1 log N log2 -1 )

Giakkoupis [STACS’11] Improved to O( -1 log N)

Page 65: Rumour Spreading in Social Networks

ResultsChierichetti, Lattanzi, P [SODA’10] With high

probability, regardless of the source, push-pull broadcasts the message within

O(log4 N/ 6) many rounds

Chierichetti, Lattanzi, P [STOC’10] Improved to O( -1 log N log2 -1 )

Giakkoupis [STACS’11] Improved to O( -1 log N)

Page 66: Rumour Spreading in Social Networks

ResultsChierichetti, Lattanzi, P [SODA’10] With high

probability, regardless of the source, push-pull broadcasts the message within

O(log4 N/ 6) many rounds

Chierichetti, Lattanzi, P [STOC’10] Improved to O( -1 log N log2 -1 )

Giakkoupis [STACS’11] Improved to O( -1 log N)

Page 67: Rumour Spreading in Social Networks

Results

Θ( -1 log N)

Page 68: Rumour Spreading in Social Networks

Variationson the theme

Dörr, Fouz, Sauerwald [STOC’11] Time for Push-Pull in PA graphs becomes O(log N/ loglog N) if random choice excludes last used neighbour

Page 69: Rumour Spreading in Social Networks

Variationson the theme

Dörr, Fouz, Sauerwald [STOC’11] Time for Push-Pull in PA graphs becomes O(log N/ loglog N) if random choice excludes last used neighbour

Page 70: Rumour Spreading in Social Networks

Variationson the theme

Dörr, Fouz, Sauerwald [STOC’11] Time for Push-Pull in PA graphs becomes O(log N/ loglog N) if random choice excludes last used neighbour

Page 71: Rumour Spreading in Social Networks

Variationson the theme

Dörr, Fouz, Sauerwald [STOC’11] Time for Push-Pull in PA graphs becomes O(log N/ loglog N) if random choice excludes last used neighbour

Page 72: Rumour Spreading in Social Networks

Variationson the theme

Dörr, Fouz, Sauerwald [STOC’11] Time for Push-Pull in PA graphs becomes O(log N/ loglog N) if random choice excludes last used neighbour

Page 73: Rumour Spreading in Social Networks

Variationson the theme

Fountoulakis, Panagiotou, Sauerwald [SODA’12] In power law graphs (Chung-Lu)

• With 2 < α < 3 O(loglog N) rounds are sufficient, with high probability, for Push-Pull to reach a (1-ε) fraction of the network, starting from a random source

• If α > 3 then Ω(logN) rounds are necessary, with high probability

Page 74: Rumour Spreading in Social Networks

Variationson the theme

Giakkoupis, Sauerwald [STOC’11] For graphs with vertex expansion at least λ Push-Pull takes

• At most O(λ log5/2 N) rounds to reach every node, with high probability

• At least Ω(λ log2 N) rounds, with positive probability

Page 75: Rumour Spreading in Social Networks

To summarize

•There is a close connection between conductance (and other expansion properties) and rumour spreading

•Since social networks enjoy high conductance, this by itself ensures that rumours will spread fast

Page 76: Rumour Spreading in Social Networks

Things to come

•Rumour spreading without the network

• Rumour spreading in evolving graphs

Page 77: Rumour Spreading in Social Networks

THANKS