Performance oriented mining of utility frequent itemsets

Abstract

The Aim of Association Rule Mining(ARM) is to find Frequent itemsets. Apriori Algorithm is one of the most efficient Frequent itemset mining Algorithm. However Frequent itemset mining does not includes interestingness or utility. Utility mining is a new area in data mining which considers all external utility factors. A specialized form of Association Rule Mining is utility-frequent itemset mining, here both utility factors and itemset frequencies are considered. Fast utility frequent itemset mining (FUFM) is one of the efficient algorithm to find utilityfrequent itemsets. The performance of an Algorithm depends on several factors like space(memory), computing time, cyclomatic complexity, external data dependency and so on. The proposed system aims in reducing the computing time of existing FUFM by implementing a Parallel computing strategy, the proposed Algorithm is Parallel implementation of Fast Utility Frequent itemset Mining algorithm(P-FUFM). Utility-frequent itemset mining algorithm consists of two phases, candidates generation and utilities generation. Utility generation is just a product function whereas candidate generation is a iterative selection process, hence the proposed algorithm is to implement parallel generation of candidate keys and standalone strategy for utilities generation. The proposed implementation and results shows that P-FUFM computes utility-frequent itemsets in very less computing time and is more suitable for Business Development.

A Kind of Authentication Method based on User Web Browsing Features

“…The former can be classified into following methods: Traditional account, password authentication [3]- [5], Smart card-based authentication [6], the authentication based on biological and behavioral characteristics (e.g. fingerprint, users' habits of using the mouse [7], [8] and keyboard input [9], [10]). Since the last one cannot be easily imitated, it attracts more academic attention.…”