A Survey on Frequent Itemset Mining in Parallel Computing Environment

Abstract

Currently there is explosive growth of information in all the fields of marketing, science, technology etc. [15]. Frequent pattern mining is the process of knowledge discovery from immense database. Number of research papers has been discovered on frequent pattern mining. Now a day frequent pattern mining on single processor or node has become bottleneck because, millions of transactions emerge as a result of large data entities. In this paper we present a survey of various frequent pattern mining algorithms which have been proposed on parallel computing environment. Parallel computing has been an efficient way for frequent pattern mining like massive computational task. We have made a survey on very important and well known papers regarding parallel frequent pattern mining. In these papers various traditional methods have been taken as base and developed a novel approach on them parallel. Apriori [1] and FP-tree [2] have been very famous and efficient frequent pattern mining algorithms. But they are not sufficient enough in this era of data mining. A new approach of Inverted matrix [7] has been discussed regarding parallel environment. This approach also overcomes various inefficiencies of the conventional approaches of Apriori and FP-growth algorithms. We have presented a table with parameters which evaluate all approaches represented here in parallel computing environment.