Mining maximal frequent itemsets by a boolean based approach

Abstract

We propose a new boolean based approach for mining frequent patterns in large transactional data sets. A data set is viewed as a truth table with an output boolean function. The value of this function is set to one if the corresponding transaction exists in the data set, zero otherwise. The output function represents a condensed form of all the transactions in the data set and is represented by an efficient compact data structure. It is then explored with a depth first strategy to mine maximal frequent itemsets. We have developed a prototype, and first experiments have shown that it is possible to rely on a condensed representation based on boolean functions to mine frequent itemsets from large databases.