reduce size of DTM to make computation faster

reduceDTM.r

reduceDTM <- function(dtm){
 term_tfidf <- 
    tapply(dtm$v/row_sums(dtm)[dtm$i], dtm$j, mean) *
    log2(nDocs(dtm)/col_sums(dtm > 0))
 
  dtm <- dtm[,term_tfidf >= median(term_tfidf)]
  dtm <- dtm[row_sums(dtm) > 0,]
  
  return(dtm)
}