On efficient top-k query processing in highly distributed environments