Categorization stuff

[Grant Ingersoll]

Hi Robin,

I have been looking a bit at the classification stuff a bit more and  
am wondering if we should be switching to use Vectors now, since the  
name could be the label and the value can contain weights, similar to  
what we do for clustering.

Also, I was wondering if you could document the format used for the  
input files now and the steps taken by the algorithms.  I'm trying to  
better understand the Wikipedia examples and also the HBase.

-Grant