Treemaps are a method for visualizing information that completely and recursively subdivide a given area into cells, where each cell’s area corresponds to certain attributes in the data set. In other words, you can look at an otherwise large set of data, and instantly distinguish trends and patterns from the wealth of information in front of you.
Take Marcos Weskamp’s Newsmap, where news headlines are aggregated from Google News, organized by subject, displayed in tightly assembled boxes, and sized according to popularity for a quick and informative look at the world’s current events.
Building on that idea are Michael Balzer and Oliver Deussen from the University of Konstanz, Germany, who use an algorithm based on Voronoi tessellations, a method which allows for organizing data into more complex shapes like shown in the image above, producing a more intuitive and flexible treemap that closely resembles the patterns formed by soap bubbles or living cells.
Check out their PDF article for the nitty-gritty math behind this and a few more images.