Så ser den alltså ut, vår kära bloggosfär.

Bilden där tagen från Data Mining: Text Mining, Visualization and Social Media. Tryck på läs vidare för att läsa om hur bilden kom till och varför den ser ut som den gör.


Data Mining förklarar:

Data - I took a hefty amount of blog data - approximately 6 weeks of full index from blogpulse. I then pulled out all the links in posts that were to other blogs and created a data set of blog to blog links.

Graph Building - By inspecting this data set for blogs that have reciprocal links (A links to B and B links to A) we can form a graph of what we might call a social network of the blogosphere.

Partitionaing - This graph will have distinct partitions. For two partitions (X and Y) there are no links between any blog in X and any blog in Y. Each partition may be thought of as a community.

Layout - Each of the communities can be laid out using a standard graph layout algorithm. Further, as there are non-reciprocal links between some of the communities we can actually use these links to layout the different communities with respect to each other (this can be thought of almost as hierarchical graph layout).

Projection - A blog is selected to be the centre of the image and the whole picture is projected on a hyperbolic surface (which gives it something like a fish-eye lens look).

What I have done in this instance is select a blog in a part of the graph that is off centre and used it to form the centre of the projection - thus pushing of the large core mass of the blogosphere to the edge of the hyperbolic surface.