We propose a strategy to perform query processing on P2P similarity search systems based on peers and superpeers. We show that by approximating global but resumed information about the indexed data in each peer, the ...