Association rule mining often generates a large number of rules, many of which may be similar, thus not containing much novel information. Design an efficient algorithm that compresses a large set of patterns into a small compact set. Discuss whether your mining method is robust under different pattern similarity definitions.