Graphing Other Things? Other ways networks can be used.

Here is an example of a graph i made which is an attempt at graphing at out Hamlet’s speech “To be or Not To Be”. In order to do so, I utilised the website textexture – a great and innovate attempt at graphing out language and text in a way that attempts to find connections between how different words are read. (read more at the website)

to be or not to be

Twitter Graphing

I decided to graph out my twitter network. Im pretty new to twitter, and dont ‘participate’ much, but rather use it as a news source/link aggregator. Here are my twitter graphs, displaying different centralities again.

DEGREE CENTRALITY

twit degree

 

 

BETWEENNESS CENTRALITY

twit betweeness

CLOSENESS CENTRALITY

twit closeness

PAGERANK

twit pagerank

 

 

Now… What can YOU tell ME about ME?

 

Visually Comparing Graph Centralities

Today I will show you  a different way to understand the differences between graph centralities.

Here is a Social Network graph i made with my class mates in one of my honours electives.

social na class a_teggin

 

Using this same data, i adjusted the node sizes according to degree; closeness; betweenness and eigenvector/PageRank.

Compare the results below, and see if you can deduce what these CENTRALITIES can indicate:

degree

betweeness

closeness

page rank

 

 

 

Http graphing with Gephi

So in experimenting different ways to graph networks online, I came accross a neat exercise you can do with Gephi and the Http Plug.

Here are two examples of networks I mapped using this software. This ego-centric map shows the different servers that are connected to my ip address when I log onto the internet. In the first example I accessed my universities library website and logged into a proxy server, while the second example shows me accessing the universities website and then accessing the department for film and media studies (my department xD)

Example 1)
uct

Example 2)
uct cfms

It is interesting to see how sites like Google analytics or wordpress float around in the background watching what we are up to.

HERE IS HOW TO DO THIS

  1. Go to this site, download Gephi for your operating system, and follow the steps to install.
  2. Once installed, open Gephi, select New Project.
  3. Select the tab Tools, and then select Plugins
  4.  Go to Available Plugins, scroll down to HttpGraph, select it and press Install.
  5. Now go get Mozilla Firefox, if you don’t already have it ( you don’t have to use Mozilla, but I find it works the smoothest for this application.)
  6. Close all your programs and restart your browser and Gephi.
  7. When you open a New Project in Gephi this time, select File > Generate > HttpGraph
  8. Keep the port as the default 8088, and select Ok
  9. Start Firefox, go to Firefox > Options > Advanced > Connection > Settings
  10. Select Manual Proxy Configuration and enter in 127.0.0.1 into the proxy and 8088 into the listening port. Select Use this proxy server for all protocols and press Ok
  11. Now you can browse through the internet and watch as the graph is made in real-time with Gephi. You can also watch as it dynamically shifts around as the networks change.
  12. Lastly, don’t forget to change firefox’s proxy servers back to normal after you are finished to continue browsing.

The Friendship Paradox & Dunbar’s Number

By analyzing large networks, some interesting theories and applications have emerged.
The friendship paradox is one of the statistical paradoxes that frequently emerges in large networks.

This article from the NY times website describes this phenomena in quite some detail. This quote shows the general idea of the friendship paradox:

For any network where some people have more friends than others, it’s a theorem that the average number of friends of friends is always greater than the average number of friends of individuals.

Dunbar’s Number is another phenomena, extracted from applied evolutionary studies. Dunbar began formulating these ideas long before the internet and social networks became popular. HE looked at various civilization’s from differently sized groups of people throughout history, eventually postulating the average size of groups/communities to be around 100 – 230 (averagely around 150)

Dunbar also looked at the habits of Christmas cards being delivered in and around London, in order to calculate the networks that emerged and track the sizes and characteristics of modern communities. Read more here

These theories are only a few interesting things that can be seen or calculated through network science and social network analysis!