Privacy-Preserving in Outsourced Transaction Databases from Association Rules Mining

Data mining-as-a-service has been selected as considerable research issue by researchers. An organization (data owner) can outsource its mining needs like resources or expertise to a third party service provider (server). However, both the association rules and the items of the outsourced transaction database are private property of data owner. The data owner encrypts its data, send data and mining queries to the server, and accepts the true patterns from the encrypted patterns received from the server to protect the privacy. The problem of outsourcing transaction database within a corporate privacy framework is studied in this paper. We propose an attack model based on previous knowledge and devise a scheme for privacy preserving outsourced data mining. Our scheme ensures that each transformed data is different with respect to the attacker’s previous information. The experimental results on real transaction database prove that our techniques are scalable, efficient and protect privacy. Index Terms — Privacy-preserving outsourcing, Association rule mining