Week 7 Reflections

Gephi, my Gephi

This week we learned about a network visualization tool called Gephi (http://www.gephi.org). This is a cool open source tool that provides a set of powerful analysis tools for networks. Of course, since it is open-source, that comes with its own issues. The current version is a beta release, and there are still some glitches, like screen refreshes that don't happen, or randomly changing colors when you re-run a graph (ok, this only bugs me because I am trying to take screenshots). But it's free, so who am I to complain? Overall, it's pretty good software, although the help/tutorials need some serious help.

Next we looked at netvizz (https://tools.digitalmethods.net/netvizz/facebook), which is a Facebook application that analyzes one's Facebook friend network. Upon completing the analysis, the tool produces a GDF file which can be imported into Gephi! We looked at the Personal Network analysis, although netvizz allow you to analyze groups and pages as well. The tool produces the following data:

file fields (network file - gdf format - nodes are users):
sex: user specified sex
locale: user selected interface language
agerank: accounts ranked by creation date where 1 is youngest
like_count: number of user likes
post_count: number of user posts
post_like_count: number of likes on user's posts
post_comment_count: number of comments on user's posts
post_engagement_count: post_comment_count + post_like_count

When the network was first imported, it did remind me a lot of the Borg cube:

I guess I could go on with Facebook/Borg comparisons, but probably shouldn't...

What's interesting is that during the analysis, I found several discrete clusters:

It's interesting to actually see this information... you wouldn't really understand this without visualization.

It was also interesting to zoom in and see who the influential people in my network are. I guess if I'm planning on spreading some rumors, I know who to tell first!

I think this could also be useful when deciding who to "defriend" on Facebook. If the goal is to keep important people in one's network, then knowing who those people are (via analysis tools like Gephi) can be very helpful indeed!