Have you got a -cl option in your command line? All the clustering
routines now observe it so that the step of clustering the points is
uniformly optional
On 5/23/10 6:02 PM, Delroy Cameron wrote:
so in kmeans clustering, shouldn't the points directory be created when the
vectors are created, as is the case with the dictionary,
tokenized-documents, wordcount etc.
since i don't have a points directory, also i don't have any points in the
clusterdump.
instead the clusterdump has only the terms, their scores and top-k terms.. i
need to know the docs in each cluster as well..
how can i get the points/docs in each cluster?
-----
--cheers
Delroy
