Privacy Preserving Mining Association Rule from Outsourced Transaction

Abstract

Reproached by developments such as cloud computing, there has been considerable recent interest in paradigm of data mining-as-a-service. The company (data owner) lacking in expertise or computational resources can outsource it’s to the third revelry service worker (server). However, both items and association rules of outsourced database are considered private property of corporation (data owner). To protect corporate privacy, data owner transforms its data and ships it to server, sends mining queries to server, and recovers true patterns from extracted patterns received from server. In these paper, we study problem of outsourcing association rule mining task within the corporate privacy-preserving outline. We recommend an attack model based on background knowledge and devise the scheme for privacy preserving outsourced mining. Our scheme ensures that each transformed item was indistinguishable, w.r.t. attacker’s background knowledge, from at minimum k-1 other transformed items. Our comprehensive experiments on the very large and real transaction database establish that our systems are effective, scalable, and protect privacy.