I wanted to learn more about the power of visualising these networks by examining clustering of my Facebook network. For obvious privacy issues, I have removed the labels on the nodes (sorry!). I simply want to present and discuss the structures. Try this stuff out on your own, you’ll be surprised with what you find out. T-shirt with this design is now available on Redbubble.com.
A lot of this you’ll just have to take my word on. In the top left is a rather large cluster of red nodes. I had a look at the names associated with these nodes and found that all of these people were friends from high school. The large blue blob are friends from college. The yellow nodes are my girlfriend’s friends which forms its own sub-network as one would expect. In the bottom left we have my university colleagues and surprisingly in the very bottom left, as an offshoot to my university colleagues, we have academic colleagues who aren’t at my university. If you spend a few seconds thinking about how friendships are likely to be made, it will become clear why the network takes on this structure. Lastly, there are a few groups not located near any other nodes. This is because these are groups of friends I’ve made whilst travelling or at conferences. Those friends have their own network and are highly likely not to have any connections to anyone else in the network.
We can also filter out the oldest people (or youngest) people in the network:
We can also filter according to gender. Males only:
Two interesting things here:
a) My girlfriend’s network almost entirely disappears when the male filter is turned on. This is probably because there are few male friends of my girlfriend’s friends with whom I am friends. I’m not sure whose fault that is .
b) When the female filter is turned on, my university friendship network dissapears because I work in a male dominated environment.
You can even sort by wall count and see who are the busiest social networkers in your group but I digress.
So how can you do it?
1. Type “netvizz” in your Facebook profile. Enable this app and create the file required to be read by Gephi.
2. Load into Gephi and try the Force Atlas 2 algorithm and enture no overlap and dissuade hub is selected.
3. Run this for about 10 seconds and boom, you have a visual representation of your social network (with some minor extra tinkering).