A Kernel Based Apriori Algorithm for Sequential Pattern Mining

Abstract

Several researchers focused on the sequential pattern mining problem and many algorithms weredeveloped to mine sequential patterns. In this paper, we propose an efficient algorithm K-Apriori which makes use of a new pruning methodcalled Hamming Distance that allows the early detection of sequential patterns during themining process. This approach is an extension to the most popular Apriori algorithm.  This K-Apriori algorithm forms the kernels based on the membership of each items and then grouped them into one kernel those having same membership values. Since, the items are grouped; there is no need to scan all the items for every mining process. This approach reduces the time taken for mining greatly. The extensive simulations reveal the performance of proposed approach under various test cases.