Privacy Preserving Fuzzy Association Rule Mining in Data Clusters Using Particle Swarm Optimization

Abstract

An association rule is classified as sensitive if its thread of revelation is above certain confidence value. If these sensitive rules were revealed to the public, it is possible to deduce sensitive knowledge from the published data and offers benefit for the business competitors. Earlier studies in privacy preserving association rule mining focus on binary data and has more side effects. But in practical applications the transactions contain the purchased quantities of the items. Hence preserving privacy of quantitative data is essential. The main goal of the proposed system is to hide a group of interesting patterns which contains sensitive knowledge such that modifications have minimum side effects like lost rules, ghost rules, and number of modifications. The proposed system applies Particle Swarm Optimization to a few clusters of particles thus reducing the number of modification. Experimental results demonstrate that the proposed approach is efficient in terms of lost rules, number of modifications, hiding failure with complete avoidance of ghost rules.