Mining Generalized Closed Frequent Itemsets of Generalized Association Rules

Abstract

In the area of knowledge discovery in databases, the generalized association rule mining is an extension from the traditional association rule mining by given a database and taxonomy over the items in database. More initiative and informative knowledge can be discovered. In this work, we propose a novel approach of generalized closed itemsets. A smaller set of generalized closed itemsets can be the representative of a larger set of generalized itemsets. We also present an algorithm, called cSET, to mine only a small set of generalized closed frequent itemsets following some constraints and conditional properties. By a number of experiments, the cSET algorithm outperforms the traditional approaches of mining generalized frequent itemsets by an order of magnitude when the database is dense, especially in real datasets, and the minimum support is low.