This is equivalent to projecting all data points onto a sphere and then using a symmetric Gaussian and is an excellent choice for many problems.
On Wed, Nov 2, 2011 at 2:29 PM, Jeff Eastman <[email protected]> wrote: > I've updated build-reuters.sh to use a CosineDistanceMeasure and a > DistanceMeasureCluster instead and the performance has improved over 100x > on Reuters.
