Mining essential rules using frequent closed itemsets

Abstract

According to the traditional association rules mining, finding all association rules satisfied minSup and minConf will face to many disadvantages in cases of the large frequent itemsets. Thus, it is necessary to have a suitable method for mining in number of fewer rules but make sure fully integrating rules of traditional methods. In this paper, we present an algorithm of generating essential rules from frequent closed itemsets: only stores rules having smallest antecedent and largest consequent based on parents-child relationship. Experiment shows that the resulted rule set is rarely as small as traditional set, the time for rule mining is also faster than the time for traditional because the mining essential rules are based on frequent closed itemsets (FCI) whereas mining traditional rules based on frequent itemsets (FI) that satisfies |FCI|  |FI|.