Iterative Algorithm for Finding Frequent Patterns in Transactional Databases

Abstract

A high-performance algorithm for searching for frequent patterns (FPs) in transactional databases is presented. The search for FPs is carried out by using an iterative sieve algorithm by computing the set of enclosed cycles. In each inner cycle of level FPs composed of elements are generated. The assigned number of enclosed cycles (the parameter of the problem) defines the maximum length of the desired FPs. The efficiency of the algorithm is produced by (i) the extremely simple logical searching scheme, (ii) the avoidance of recursive procedures, and (iii) the usage of only one-dimensional arrays of integers.