Archimedean Spiral for CorpusClusterView/TextClusterView

[jarmoza]

Since the distance from the average topic distributions of these views remains a 1-dimensional distance (as established by the Jensen-Shannon probability distribution distance calculation), one proposed visual of this is an Archimedean spiral, with topic bullseyes/rectangles emanating out from the center at their calculated Jensen-Shannon distance.

This not only solves the arbitrariness of the 2nd dimensional coordinate as calculated for the force-directed graph, but can also eliminate node overlap and confusion as to comparative distances of those nodes from the center of the graph/average topic distribution. This should also eliminate slowdown caused by the calculation of static, non-colliding positions of bullseyes/rectangles in the current force-directed graph.

An implementation of this spiral in D3 (albeit with equidistant points) can be seen here at the bottom of the page:

http://stackoverflow.com/questions/27596115/finding-x-y-coordinates-of-a-point-on-an-archimedean-spiral

Possible animation example (albeit a fibonacci spiral):

http://www.bytemuse.com/post/fibonacci-spiral/