I'm not sure about modules for the data reduction phase, but here are a some hints to head you along the way:

1. Create a hash by year of lists of item id and count pairs (perlref and perllol may help)
2. for each year sort the appropriate list by count (see sort for help)
3. Use an array slice to pick off the top 10 items (see the slices section in perldata)