The visualisation of complex multidimensional relationships as geographic maps is a powerful way to combine notions of clustering and topological relationships in an intuitive and playful way.
Bookland is a good example of how the automatic clustering and then mapping of relationships can have real significance.
A visualization of the TVLand using our proposed technique. The data comes from a digital TV service with over a million set top boxes - arxiv.org 
TVLand visualizes data from a digital TV service with over a million set top boxes. For the top 1000 most popular shows, we take the top 10 most similar shows, which gives us a sparse matrix.
Instead of one monolithic cluster of news shows, our map shows several distinct sub-clusters.
Leftbank Newsistan anchored by classic CNN shows Newsroom, The Situation Room and Larry King.
In the SW of the continent is the compact and well interconnected Leftbank Newsistan anchored by classic CNN shows Newsroom, The Situation Room and Larry King along with MSNBC newcomers Countdown with Keith Olbermann and The Rachel Maddow Show and the off-kilter Comedy Central shows Daily Show with Jon Stewart and Colbert Report.
Rightbank Newsistan anchored by Fox and Friends, Hannity, and The O'Reilly Factor
Diametrically opposite on the East is the similarly compact and well interconnected Rightbank Newsistan anchored by Fox and Friends, Hannity, and The O'Reilly Factor
Above are a couple of clusters of local news and below is yet another news-cluster, mostly made up of morning news shows.
It is worth noting that the seemingly meaningful left-right placement of the two distinct news clusters was co-incidental. However, the diametrically opposing placement of these two clusters is meaningful as, although they both contain news-related shows, there are very few viewers who regularly watch shows in both clusters.
High-dimensional data are visualized as a collection of points in 2-dimensional space through dimensionality reduction techniques - pdf
GvMap is an algorithm for visualizing graphs as maps. It is part of the Graphviz library, and was originally called the GMap Algorithm.