Privacy Preserving Data Mining on Big Data Computing Platform: Trends and Future

Abstract

Data mining is becoming increasingly important in the data-driven society in recent years. Unfortunately, privacy of the individuals fails to be protected and considered deliberately. It’s a significantly challenging question that outputs of data mining models can be applied to preserve privacy while simultaneously maintaining analyzing capability. With advancements in big data, series of big data computing platforms have evolved into widely utilized paradigms for data mining. However, users’ sensitive data which are outsourced on the cloud and mined on open-sourced computing platform. It poses such severe threats that measures must be taken to protect the privacy of individuals’ data. Regarding this issue, much fruitful work has been done on designing privacy preserving data mining approaches for improving big data computing platform security and privacy of individuals. In this paper, a systematic investigation of a wide array of the state-of-the-art privacy preserving data mining (PPDM) techniques has been performed from different aspects on threat model, anonymity, secure multiparty computation (SMC), differential privacy. We are focused on improving data privacy in these sensitive areas on big data computing platforms. Hopefully, our work aims to highlight the urgent need for applying privacy preserving data mining approaches on big data computing platforms. Moreover, a better understanding of this research area may benefit the usage of big data and future exploration.