Hi, just wanted to try a svm on the keywords to find significant keywords, for example for US users ( http://havard.security-review.net/Bittorent.pdf ). But first had to find a ideal approach to pre-process the data in R -Håvard