splitreuters21578.sh

#!/bin/bash                                                                                                 

#This script simply splits up the reuters-21578 dataset into separate files for each article.

for f in reut2-*.sgm
do
    echo $f
    sed '1d' $f | csplit -ks -n 3 -f split/${f%.sgm} - '/<REUTERS/' {100000}
done