Fast Algorithms for Mining Interesting Frequent Itemsets

Abstract

Certifiable datasets are inadequate, dirty and contain hundreds of things. In such situations, discovering interesting standards (comes about) using traditional frequent itemset mining approach by specifying a client defined input bolster edge is not fitting. Since with no domain knowledge, setting bolster edge little or substantial can yield nothing or a huge number of redundant uninteresting outcomes. Recently a novel approach of mining only N-most/Top-K interesting frequent itemsets has been proposed, which finds the top N interesting outcomes without specifying any client defined help edge. Be that as it may, mining interesting frequent itemsets without minimum help limit are all the more exorbitant as far as itemset look space exploration and processing cost. In this way, the efficiency of their mining profoundly depends upon three main components (1) Database representation approach utilized for itemset frequency counting, (2) Projection of relevant transactions to bring down level nodes of inquiry space and (3) Algorithm implementation technique. Thusly, to enhance the efficiency of mining process, in this paper we present two novel algorithms called (N-MostMiner and Top-K-Miner) using the bit-vector representation approach which is extremely efficient as far as itemset frequency counting and transactions projection. In addition to this, few efficient implementation techniques of N-MostMiner and Top-K-Miner are additionally present which we experienced in our implementation. Our experimental outcomes on benchmark datasets propose that the N-MostMiner and Top-K-Miner are extremely efficient as far as processing time when contrasted with current best algorithms BOMO and TFP.