An Optimized Sanitization Approach for Minable Data Publication

Minable data publication is ubiquitous since it beneficial to sharing/trading among commercial companies and further facilitates the development of data-driven tasks. Unfortunately, minable often implemented by publishers with limited privacy concerns such that published dataset malicious entities. It prohibits may contain sensitive information. Thus, urgently demanded present some approaches technologies for reducing leakage risks. To this end, in paper, we propose an optimized sanitization approach (named as SA-MDP). SA-MDP supports association rules mining function while providing protection specific rules. In SA-MDP, consider trade-off between utility problem. address problem, designs a customized particle swarm optimization (PSO) algorithm, where objective determined both privacy. Specifically, take advantage PSO produce new particles, which achieved random mutation or learning from best particle. Hence, can avoid solutions being trapped into local optima. Besides, design proper fitness guide particles run towards optimal solution. Additionally, preprocessing method before evolution process algorithm improve convergence rate. Finally, proposed performed verified over several datasets. The experimental results have demonstrated effectiveness efficiency SA-MDP.