Scribd
Upload
24 views

Mining Social - Network Graph

This document discusses techniques for mining social network graphs. It describes what a social network is - a collection of entities with relationships between them where related entities are more likely to be related. It presents social networks as graphs that can be partitioned into communities. Methods discussed for graph partitioning include normalized cut and laplacian eigenmaps. Laplacian eigenmaps work by mapping the graph to vectors such that connected components stay close together.

Uploaded by

Ahad Haider
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
24 views

Mining Social - Network Graph

This document discusses techniques for mining social network graphs. It describes what a social network is - a collection of entities with relationships between them where related entities are more likely to be related. It presents social networks as graphs that can be partitioned into communities. Methods discussed for graph partitioning include normalized cut and laplacian eigenmaps. Laplacian eigenmaps work by mapping the graph to vectors such that connected components stay close together.

Uploaded by

Ahad Haider
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Mining

Social-Network Graph

What is Social Network?


There is a collec9on of en99es that par9cipate
in the network
There is at least one rela9onship between
en99es
There is an assump9on of nonrandomness or
locality, i.e if en9ty A is related to both B and
C , then there is a higher probability than
average that B and C are related

Social Network as Graph

Graph Par99oning
What makes a good par99on

Normalized Cut

Normalized Cut - Example

Laplacian Eigenmap
How do we map the graph to a line, so that connected components stay as
close to each other as possible.

Algorithmic steps:
1. Create Adjacency matrix, A

Laplacian Eigenmap
2. Create degree matrix, D

Adjacency matrix

3. Create Laplacian Matrix, L = D-A

Op9mal mapping is provided by the matrix of eigenvectors


corresponding to the lowest eigenvalues of the generalized
eigenvalue problem
Ly = Dy

Example

You might also like