It is great to see your question, i would greatly appreciate it.

Here is the answer for you.

Code:

ls -lh
total 28K
-rw------- 1 thegeek learner  18K Oct 10 05:52 feed
-rw------- 1 thegeek learner 2.5K Oct 10 05:51 feedbacks
-rw------- 1 thegeek learner   32 Oct 10 05:51 tttt

So only the feed is the file whose size is 18k which is more than the half of the dir size which is 28K.

To get that, following is the script.,

Code:

for i in `find . -type d`
do 
    echo $i
    dirsize=`du -s $i | cut -f1`
    echo $dirsize
    find $i -size +$dirsize -type f
done

If it requires a little edit for your requirement, go a head and do it. But this would help you in achieving the core part of what you would want.