Privacy-Preservation in Collaborative Association Rule Mining for Outsourced Data

Abstract

In recent years, the explosion of digital data and information, and various applications such as real-time monitoring, distributed collaboration, large-scale medical and financial data analysis and social network accumulate large amounts of data from different data owners. The burgeoning ability to generate vast volumes of data presents technical challenges for efficient data mining. Meanwhile, with the emergence of cloud computing and its model for IT services, which affords both computational and storage scalability, the outsourcing of data for storage and mining services is acquiring popularity. So, many organizations having insuffi cient storage and computational resources, willing to reduce their storage and computation cost, are widely adopting the outsourcing of the data mining jobs to a third party service provider. These service providers are assumed to be semi-trusted parties for privacy concerns. In this paper, we propose a collaborative privacy-preserving data mining (CPPDM) solution for outsourced data, which ensures that the data is stored, processed and shared without violating the user privacy. In our solution, we are using anonymization and encryption techniques for user privacy.

Keywords: CPPDM, Outsourcing, Privacy-preservation, Semi-trusted, Encryption, Anonymization, MapReduce.

View PDF