Scribd
Upload
51 views

Community Detection in Social Networks

This document discusses community detection in social networks. It begins by introducing properties of complex networks like power laws and community structure. It then discusses why community detection is important by providing examples of how communities can represent related topics. The document compares community detection to clustering, noting key differences for network data like using graph properties directly. It presents two recent community detection methods - one based on shortest path betweenness and another based on maximizing network modularity. It provides details on calculating betweenness and modularity and discusses optimizations and extensions of these approaches.

Uploaded by

Roheet Bhatnagar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
51 views

Community Detection in Social Networks

This document discusses community detection in social networks. It begins by introducing properties of complex networks like power laws and community structure. It then discusses why community detection is important by providing examples of how communities can represent related topics. The document compares community detection to clustering, noting key differences for network data like using graph properties directly. It presents two recent community detection methods - one based on shortest path betweenness and another based on maximizing network modularity. It provides details on calculating betweenness and modularity and discusses optimizations and extensions of these approaches.

Uploaded by

Roheet Bhatnagar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 24

Community Detection in Social

Networks
Lei Tang
Properties of Complex Network

Power Law

Community Structure

Small World
Why Community Detection?
ƒ Communities in a citation network might represent
related papers on a single topic;
ƒ Communities on the web might represent pages of
related topics;
ƒ Community can be considered as a summary of
the whole network thus easy to visualize and
understand.
ƒ Sometimes, community can reveal the properties
without releasing the individual privacy information.
Community Detection,
Reinventing the wheel?
Community Detection = Clustering?

ƒ As I understand, community detection is


essentially clustering.
ƒ But why so many works on Community
Detection? (in physical review, KDD, WWW)
ƒ The network data pose challenges to
classical clustering method.
Difference
ƒ Clustering works on the distance or similarity matrix (k-
means, hierarchical clustering, spectral clustering)
ƒ Network data tends to be “discrete”, leading to algorithms
using the graph property directly (k-clique, quasi-clique,
vertex-betweenness, edge-betweeness etc.)
ƒ Real-world network is large scale! Sometimes, even n^2 in
unbearable for efficiency or space (local/distributed
clustering, network approximation, sampling method)
Outline
ƒ Two recent community detection methods
ƒ Clustering based on shortest-path
betweenness
ƒ Clustering based on network modularity
Basic Idea
ƒ A simple divisive strategy:
ƒ Repeat
1. Find out one “inter-community” edge
2. Remove the edge
3. Check if there’s any disconnected components (which
corresponds to a community)
How to measure “inter-community”
ƒ If two communities are joined by a few inter-community
edges, then all the paths from one community to another
must pass the edges.
ƒ Various measures:
ƒ Edge Betweenness: find the shortest paths
between all pairs of nodes and count how
many run along each edge.
ƒ Random Walk betweenness.
ƒ Current-flow betweenness
Shortest-path betweenness
ƒ Computation could be expensive: calculating the shortest
path between one pair is O(m), and there are O(n^2) pairs.
ƒ Could be optimized to O(mn)
ƒ Simple case: only one shortest path
When there is only one single path between the
Source S and other vertex, then those paths form
a tree.

Bottom-up: start from the leaves, assign edges to 1.

Count of parent edge = sum (count of children edge)+1


Multiple shortest path
ƒ First compute the number of paths from source to other
vertex
ƒ Then assign a proper weight for the path counts
ƒ sum of the betweenness =.number of reachable
vertices.
Calculate #shortest path
1.Initial distance

W:Number of shortest paths


Update edge weight

Edge weight
Time Complexity
ƒ O(mn) in each iteration.
ƒ Could be accelerated by noting that only the
nodes in the connected component would be
affected.
ƒ Some other techniques developed: sampling
strategy to approximate the betweenness; use
specific network index for speed.
Obtain a hierarchical tree, use modularity
To determine the number of clusters.
Modularity
ƒ Spectral clustering essentially tries to minimize the number
edges between groups.
ƒ Modularity consider the number edges which is smaller than
expected.

ƒ If the difference is significantly large, there’s a community


structure inside.
ƒ The larger, the better.
Quiz
ƒ Given a network of m edges, for two nodes
with degree ki, kj, what is the expected
edges between these two nodes?
Modularity Calculation

ƒ Modularity can be used to determine the


number of clusters, why not maximize it
directly?
ƒ Unfortunately, it’s NP-hard/
Relaxation

Eigen Value
Problem!

Modularity
Matrix

Betai is the eigen value of the


Eigen vector ui of modularity matrix B
Properties of Modularity Matrix
ƒ

ƒ (1,1,…1) is an eigen vector with zero eigen value.


ƒ Different from graph Laplacian, the eigen value of
modularity matrix could be +, 0 or -.
ƒ What if the maximum eigen value is zero?
ƒ Essentially, it hints that there’s no strong
community pattern. Not necessary to split the
network, which is a nice property.
ƒ Here, the spectral partitioning is forced to
split the network into approximately equal-
size clusters.
Extensions
ƒ Divisive clustering
ƒ K - partitioning…
Comments
ƒ I thought spectral clustering is the end of clustering. But
here a new measure Modularity is proposed and found to
be working very well, which confirms that “research is
endless”, or “no last bug”.
ƒ Since Graph Laplacian and Modularity matrix both boils
down to a eigen value problem, is there any innate
connection between these two measures?
ƒ How could it work if we apply it directly to some classic
data representation?
ƒ Extend modularity to relational data could be a promising
direction.
ƒ There could be more opportunities than “wheels” in social
computing.
ƒ Scalability is really a big issue.
References
ƒ M.E.J.Newman, Finding community structure in
networks using the eigenvectors of matrices, Phys.
Rev. , 2006
ƒ M. E. J. Newman, M. Girvan, Finding and
evaluating community structure in networks,
Phys. Rev. 2004

You might also like