Towerless (towerless) wrote in lj_nifty,

  • Music:

Friend networks graphing

I've been recently thinking about various graphical ways to represent online communities, those in LJ in particular. There is a bunch of tools that build one sort of graph or another, but none seem to provide a good view of a community the user in question is a part of. While investigating this problem I've built a prototype tool that tries to accomplish something like that; its algorithm takes 2 levels of mutual friends for a given username and cuts off those who have less than 2 friends in the cloud, thus producing something like this:

The first image is produced with the Dot algotithm and the second one - with Twopi (both from ATT's Graphwiz package). It is interesting how one can clearly see clusters of friend subgroups on the first picture, and how the second image correlates users' LJ "social status" in a given community with density - the closer you are to the center the more connected you generally are.
This post is not intended to advertise the tool (which I don't currently plan to distribute), but rather to hopefully start a discussion about interesting ways to isolate and visualize LJ user communities. Thoughts?
  • Post a new comment


    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded