Privacy Preserving in Association Rule Mining by Data Distortion Using PSO

Abstract

Association Rule Mining is one of the core data mining tasks that is used to show the associations between data items. The distribution of data from association rule mining can bring lot of advantages for research, and business teamwork.However, huge repositories of data contain secret data and sensitive patterns that must be confined before being published. We address this problem of privacy preserving association rule mining by applying data sanitization to avoid the confession of sensitive rules while maintaining data effectiveness. Particle Swarm Optimization is an artificial intelligence technique, proficient of optimizing a non-linear and multidimensional problem which typically reaches high-quality solutions efficiently while requiring negligible parametrization. To recognize the most sensitive transactions for hiding given sensitive association rules we are with Particle Swarm Optimization technique. The performance of the algorithm is validated against representative synthetic and real datasets with some performance measures.