Parallelized BSE-QP-ICOA for ARH

Abstract

The Association Rule Hiding (ARH) is a traditional method of information shielding which is about changing the real database by removing sensitive rules without altering the quality of it. Balancing Stochastic Exploration-Quality Preserving-Improved Cuckoo Optimization Algorithm for ARH (BSE-QP-ICOA for ARH) was an efficient ARH technique to sanitize the transaction database for ARH. In BSE-QP-ICOA for ARH, a meta-heuristic algorithm called cuckoo search algorithm was used where each cuckoo inserted or deleted the sensitive items based on the multi-objective function for ARH. Because of the enormous volume of data, the implementation of BSE-QP-ICOA for ARH in a single node is no longer useful and it leads to high computational cost issue. It is required to solve these problems by using a distributed architecture. In this paper, MapReduce framework is used to solve the above mentioned problems in BSE-QP-ICOA for ARH. Initially, the transaction database is split into number of independent chunks and it is given as input to map function. In the map function, minimum number transactions for modifications are chosen and its sensitive items are altered depending on the fitness function of each cuckoo. Finally, the map function returns the fitness function of each cuckoo and it is given as input to the reducer function which chooses the best fitness function. According to the finest fitness function, the cuckoo in the map function migrate their position and sanitizes the transaction database. The whole process is named as Parallelized BSE-QP-ICOA for ARH (PBSE-QP-ICOA). Thus,y parallelizing the BSE-QP-ICOA for ARH process using MapReduce the computational cost
for ARH is reduced effectively.