1 of 8

Handling Large Networks

2 of 8

Analyze social media

  • But billions of users
  • Too big to analyze!
  • Too big to visualize!
  • Filtering or sampling might be important

3 of 8

Snowball Sampling

  • When working with a large network, choose a start node
  • Get that node, its connections, their connections, and so on
  • Until you get enough nodes to analysis
  • Problems?

4 of 8

Snowball Sampling

  • When working with a large network, choose a start node
  • Get that node, its connections, their connections, and ao on
  • Until you get enough nodes to analysis
  • Problems?
    • Biased towards that part of the network
    • May miss other features
  • Benefits
    • Easy to do
    • common

5 of 8

Random Sampling

  • Randomly select a certain percentage of nodes and keep all edges between them OR
  • Randomly select a certain percentage of edges and keep all nodes that are connected by those edges

6 of 8

Random Sampling

  • Randomly select a certain percentage of nodes and keep all edges between them OR
  • Randomly select a certain percentage of edges and keep all nodes that are connected by those edges
  • Problems
    • Edge sampling biased towards high degree nodes
      • Justin Beiber will have high probability of being selected
    • Node sampling loses structural characteristics
      • What if we miss a hub node? Everything is random after all!

​

7 of 8

Random Sampling

  • Randomly select a certain percentage of nodes and keep all edges between them OR
  • Randomly select a certain percentage of edges and keep all nodes that are connected by those edges
  • Benefits
    • Easy to do

​

8 of 8

Egocentric Network Analysis

  • Instead of looking at the whole network, look at the egocentric network of some nodes
  • A different type of analysis
  • Not the whole network analysis, but rather cluster analysis
  • A node and its environ in the network
    • Analyze that
    • Keep doing that for a bunch of nodes
  • See the roles of the nodes

​