A few weeks ago, I prepared a series of data visualizations that would explain what the interest graph is and how it relates to the social graph. The concept I came up with was very straightforward. First, plot the current individual at the center of a canvas and arc edges to their interests based on attachment/polarity. Second, overlay friends from their surrounding social network around the interest graph. Finally, display the connections between appropriate interests and friends.
This was one of the first graphs to be generated. In this particular case, it's pretty clear to see how people you know are connected to things that you are interested in.
Here's another interest graph that has substantially more nodes. As a result, the social/interest edges create some very neat color combinations.
Perhaps the most gratifying part of this process was the ability to get many of these visualizations printed as large posters. Having EPS as the output format allowed for crisp rasterization at a large size. The above was taken in the printer's warehouse.
Technically, I coded the entire process just using Core Image. Tangent: the more I use Core Image the more I'm delighted by its abstractions. The underlying data for these visualizations was provided by our currently unreleased platform. Once our APIs have been made public, you'll be able to create anything from visualizations to applications that take advantage of our interest service. I'm excited about the platform - you should be too.