Models of Network Formation

Slides:



Advertisements
Similar presentations
Peer-to-Peer and Social Networks Power law graphs Small world graphs.
Advertisements

Scale Free Networks.
Algorithmic and Economic Aspects of Networks Nicole Immorlica.
Lecture 21 Network evolution Slides are modified from Jurij Leskovec, Jon Kleinberg and Christos Faloutsos.
Synopsis of “Emergence of Scaling in Random Networks”* *Albert-Laszlo Barabasi and Reka Albert, Science, Vol 286, 15 October 1999 Presentation for ENGS.
Models of Network Formation Networked Life NETS 112 Fall 2013 Prof. Michael Kearns.
Generative Models for the Web Graph José Rolim. Aim Reproduce emergent properties: –Distribution site size –Connectivity of the Web –Power law distriubutions.
Universal Network Structure and Generative Models Networked Life CIS 112 Spring 2010 Prof. Michael Kearns.
Information Networks Small World Networks Lecture 5.
4. PREFERENTIAL ATTACHMENT The rich gets richer. Empirical evidences Many large networks are scale free The degree distribution has a power-law behavior.
CSE 522 – Algorithmic and Economic Aspects of the Internet Instructors: Nicole Immorlica Mohammad Mahdian.
1 Evolution of Networks Notes from Lectures of J.Mendes CNR, Pisa, Italy, December 2007 Eva Jaho Advanced Networking Research Group National and Kapodistrian.
Network Models Social Media Mining. 2 Measures and Metrics 2 Social Media Mining Network Models Why should I use network models? In may 2011, Facebook.
Scale-free networks Péter Kómár Statistical physics seminar 07/10/2008.
Long Tails and Navigation Networked Life CSE 112 Spring 2007 Prof. Michael Kearns.
Strategic Models of Network Formation Networked Life CIS 112 Spring 2010 Prof. Michael Kearns.
Peer-to-Peer and Grid Computing Exercise Session 3 (TUD Student Use Only) ‏
Advanced Topics in Data Mining Special focus: Social Networks.
Economic Models of Network Formation Networked Life CIS 112 Spring 2008 Prof. Michael Kearns.
NetworkNetwork Science: “Universal” Structure and Models of FormationScience Networked Life CIS 112 Spring 2009 Prof. Michael Kearns.
1 Algorithms for Large Data Sets Ziv Bar-Yossef Lecture 7 May 14, 2006
The Erdös-Rényi models
(Social) Networks Analysis III Prof. Dr. Daning Hu Department of Informatics University of Zurich Oct 16th, 2012.
Topic 13 Network Models Credits: C. Faloutsos and J. Leskovec Tutorial
Author: M.E.J. Newman Presenter: Guoliang Liu Date:5/4/2012.
Contagion in Networks Networked Life NETS 112 Fall 2013 Prof. Michael Kearns.
Analysis of Social Media MLD , LTI William Cohen
Social Network Analysis Prof. Dr. Daning Hu Department of Informatics University of Zurich Mar 5th, 2013.
Lecture 10: Network models CS 765: Complex Networks Slides are modified from Networks: Theory and Application by Lada Adamic.
Most of contents are provided by the website Network Models TJTSD66: Advanced Topics in Social Media (Social.
How Do “Real” Networks Look?
Navigation in Networks, Revisited Networked Life MKSE 112 Fall 2012 Prof. Michael Kearns.
The simultaneous evolution of author and paper networks
Contagion in Networks Networked Life NETS 112 Fall 2015 Prof. Michael Kearns.
Network (graph) Models
Lecture 23: Structure of Networks
Connectivity and the Small World
Hiroki Sayama NECSI Summer School 2008 Week 2: Complex Systems Modeling and Networks Network Models Hiroki Sayama
Trading in Networks: I. Model
Lecture 1: Complex Networks
Topics In Social Computing (67810)
How Do “Real” Networks Look?
Lecture 23: Structure of Networks
Networked Life NETS 112 Fall 2018 Prof. Michael Kearns
Random Graph Models of large networks
The Watts-Strogatz model
How Do “Real” Networks Look?
How Do “Real” Networks Look?
Social Network Analysis
Lecture 13 Network evolution
Models of Network Formation
Peer-to-Peer and Social Networks Fall 2017
Networked Life NETS 112 Fall 2017 Prof. Michael Kearns
Models of Network Formation
How Do “Real” Networks Look?
Networked Life NETS 112 Fall 2014 Prof. Michael Kearns
Networked Life NETS 112 Fall 2016 Prof. Michael Kearns
Models of Network Formation
Peer-to-Peer and Social Networks
From the Synthetic networks slide deck (didn’t have a chance to go over them at that time) By: Ralucca Gera, NPS.
Lecture 23: Structure of Networks
Lecture 9: Network models CS 765: Complex Networks
Lecture 21 Network evolution
Modelling and Searching Networks Lecture 6 – PA models
Network Science: A Short Introduction i3 Workshop
Network Models Michael Goodrich Some slides adapted from:
Advanced Topics in Data Mining Special focus: Social Networks
Discrete Mathematics and its Applications Lecture 6 – PA models
Advanced Topics in Data Mining Special focus: Social Networks
Networked Life NETS 112 Fall 2019 Prof. Michael Kearns
Presentation transcript:

Models of Network Formation Networked Life NETS 112 Fall 2017 Prof. Michael Kearns

Roadmap Recently: typical large-scale social and other networks exhibit: giant component with small diameter sparsity heavy-tailed degree distributions high clustering coefficient These are empirical phenomena What could “explain” them? One form of explanation: simple models for network formation or growth that give rise to these structural properties Next several lectures: Erdös-Renyi (random graph) model “Small Worlds” models Preferential Attachment Discussion of structure exhibited (or not) by each

Models of Network Formation I. The Erdös-Renyi (Random Graph) Model

The Erdös-Renyi (Random Graph) Model Really a randomized algorithm for generating networks Begin with N isolated vertices, no edges Add edges gradually, one at a time Randomly select two vertices not already neighbors, add edge So edges are added in a random, unbiased fashion About the simplest (dumbest?) formation model possible But what can it already explain?

The Erdös-Renyi (Random Graph) Model After adding E edges, edge density is As E increases, p goes from 0 to 1 Q: What are the likely structural properties at density p? e.g. as p = 0  1, small diameter occurs; single connected component At what values of p do “natural” structures emerge? We will see: many natural and interesting properties arise at rather “small” p furthermore, they arise very suddenly (tipping/threshold) Let’s examine the Erdös-Renyi simulator

Why Can’t There Be Two Large Components? densely connected densely connected missing edges

Threshold Phenomena in Erdös-Renyi Theorem: In Erdös-Renyi, as N becomes large: If p < 1/N, probability of a giant component (e.g. 50% of vertices) goes to 0 If p > 1/N, probability of a giant component goes to 1, and all other components will have size at most log(N) Note: at edge density p, expected/average degree is p(N-1) ~ pN So at p ~ 1/N, average degree is ~ 1: incredibly sparse So model “explains” giant components in real networks General “tipping point” at edge density q (may depend on N): If p < q, probability of property goes to 0 as N becomes large If p > q, probability of property goes to 1 as N becomes large For example, could examine property “diameter 6 or less”

Threshold Phenomena in Erdös-Renyi Theorem: In Erdös-Renyi, as N becomes large: Threshold at for diameter 6. Note: degrees growing (slightly) with N If N = 300M (U.S. population) then average degree pN ~ 500 If N = 7BN (world population) then average degree pN ~ 1000 Not unreasonable figures… At p not too far from 1/N, get strong connectivity Very efficient use of edges

Threshold Phenomena in Erdös-Renyi In fact: Any monotone property of networks exhibits a threshold phenomenon in Erdös-Renyi monotone: property continues to hold if you add edges to the networks e.g. network has a group of K vertices with at least 71% neighbors e.g. network has a cycle of at least K vertices Tipping is the rule, not the exception

What Doesn’t the Model Explain? Erdös-Renyi explains giant component and small diameter But: degree distribution not heavy-tailed; exponential decay from mean (Poisson) clustering coefficient is *exactly* p To explain these, we’ll need richer models with greater realism

Models of Network Formation II. Clustering Models

Roadmap So far: Erdös-Renyi exhibits small diameter, giant connected component Does not exhibit high edge clustering or heavy-tailed degree distributions Next: network formation models yielding high clustering Will also get small diameter “for free” Two different approaches: “program” or “bake” high clustering into the model balance “local” or “geographic” connectivity with long-distance edges

“Programming” Clustering Erdös-Renyi: global/background edge density p all edges appear independently with probability p no bias towards connecting friends of friends (distance 2)  no high clustering But in real networks, such biases often exist: people introduce their friends to each other people with common friends may share interests (homophily) So natural to consider a model in which: the more common neighbors two vertices share, the more likely they are to connect still some “background” probability of connecting still selecting edges randomly, but now with a bias towards friends of friends

Making it More Precise: the a-model y = probability of connecting u & v x = number of current common neighbors of u & v 1.0 network size N smaller a a = 1 larger a “default” probability p

From D. Watts, “Small Worlds”

Clustering Coefficient Example 2 Network: simple cycle + edges to vertices 2 hops away on cycle By symmetry, all vertices have the same clustering coefficient Clustering coefficient of a vertex v: Degree of v is 4, so the number of possible edges between pairs of neighbors of v is 4 x 3/2 = 6 How many pairs of v’s neighbors actually are connected? 3 --- the two clockwise neighbors, the two counterclockwise, and the immediate cycle neighbors So the c.c. of v is 3/6 = ½ Compare to overall edge density: Total number of edges = 2N Edge density p = 2N/(N(N-1)/2) ~ 4/N As N becomes large, ½ >> 4/N So this cyclical network is highly clustered

An Alternative Model A different model: start with all vertices arranged on a ring or cycle (or a grid) connect each vertex to all others that are within k cycle steps with probability q, rewire each local connection to a random vertex Initial cyclical structure models “local” or “geographic” connectivity Long-distance rewiring models “long-distance” connectivity q=0: high clustering, high diameter q=1: low clustering, low diameter (~ Erdös-Renyi) Again is a “magic range” of q where we get both high clustering and low diameter Let’s look at this demo

Summary Two rather different ways of getting high clustering, low diameter: bias connectivity towards shared friendships mix local and long-distance connectivity Both models require proper “tuning” to achieve simultaneously Both a bit more realistic than Erdös-Renyi Neither model exhibits heavy-tailed degree distributions

Models of Network Formation III. Preferential Attachment

Rich-Get-Richer Processes Processes in which the more someone has of something, the more likely they are to get more of it Examples: the more friends you have, the easier it is to make more the more business a firm has, the easier it is to win more the more people there are at a nightclub, the more who want to go Such processes will amplify inequality One simple and general model: if you have amount x of something, the probability you get more is proportional to x so if you have twice as much as me, you’re twice as likely to get more Generally leads to heavy-tailed distributions (power laws) Let’s look at a simple “nightclub” demo…

Preferential Attachment Start with two vertices connected by an edge At each step, add one new vertex v with one edge back to previous vertices Probability a previously added vertex u receives the new edge from v is proportional to the (current) degree of u more precisely, probability u gets the edge = (current degree of u)/(sum of all current degrees) Vertices with high degree are likely to get even more links! …just like the crowded nightclub Generates a power law distribution of degrees Variation: each new vertex initially gets k edges Here’s another demo

Summary Now have provided network formation models exhibiting each of the universal structure arising in real-world networks Often got more than one property at a time: Erdös-Renyi: giant component, small diameter α model, local+long-distance: high clustering, small diameter Preferential Attachment: heavy-tailed degree distribution, small diameter Can we achieve all of them simultaneously? Probably: mix together aspects of all the models Won’t be as simple and appealing, though