[eclipse-clp-users] Knappsack-Problem Modification

Hi,

I have a problem that can be abstracted as a modification of the 
knappsack problem. I need help to solve it.

Imagine:
Given a set of items I that have a meassure of dissimilarity amongst 
eachother: dis(Item1,Item2,Dissimilarity)
Given a knappsack K, pack items from I into K so that the average 
dissimilarity amongst the packed items is minimized, to exclude trivial 
solutions one might state a constraint that the knappsack has to contain 
at least k elements.

Later I'd like to expand this example to n knappsacks, whereby items 
should be allocated in such a way, that every item is packed to a 
knappsack and each knappsack has the lowest average dissimilarity possible.

I would be really happy if someone could give me suggestions on how to 
solve it best.

Best regards,
Philipp Marcus