Injecting Differential Privacy in Rules Extraction of Rough Set

Abstract

Data mining plays a very important role in various database applications. Medical data mining has been a popular data mining application with a vital role in improving the quality of medical services and promoting the development of the medical industry. There has been extensive research in rough set theory (RST) to mine potential patterns in medical data, which has important implications for clinical decision support and online medical diagnosis. Although medical data mining is very promising, the rapid development of this field still faces many challenges, such as information security and privacy issues. Under the assumption that data miners cannot be trusted, this paper combines the differential privacy and rough set rules’ extraction for the first time and proposes a new method to mine hidden patterns in medical data and ensure patient privacy. This algorithm uses the Laplacian mechanism to add noise to the credibility in the process of data mining while maximizing the utility of the data. Experiments show that our algorithm can effectively preserve the accuracy of data while protecting patient privacy.