Thursday, October 27, 2011

10/25/11

Rocchio Classification for text data uses existing
data clusters (ie training data) and then puts new
doc in its closest cluster (nearest centroid). 
But it does not change the centroid as is done in 
k-means.  
M.