A Study On Differential Privacy Protection Method Based On Multiple Data Sources

With the popularity of the Internet and the rapid development of mobile computation technology,data collection and utilization have become more and more convenient.While people enjoy the convenience of data analysis and application,there is also the risk of personal privacy leakage.The leakage of personal privacy has caused some troubles and even harms to people's work and life,which has aroused widespread concern in society.More and more researchers have conducted a lot of research on privacy protection.At present,most privacy protection methods protect information from a single data source.However,with the data sharing and deep application of data,more and more data of multiple data sources is being used,and there may be a certain relationship between data from different data sources.Even if the user data sensitive information in a single data source is protected,the attacker may use the corresponding methods to infer the user's sensitive information through the data association relationship between the multiple data sources.Therefore,how to prevent user privacy leakage under the application of multiple data sources has become an urgent problem to be solved.In this paper,the risk of privacy leakage in data from multiple data sources is analyzed.According to the size of non-sensitive attribute set and sensitive attribute set,NSRR-DP and SRR-DP algorithms are proposed to protect user privacy.They randomly respond to non-sensitive and sensitive attributes.They all effectively protect the privacy of the user.The specific research contents are as follows:(1)Analyzing the risk of data privacy leakage in multiple data sources,describing the privacy protection problem of multiple data sources formally,and proposing the NS-SC algorithm to calculate the non-sensitive coverage of all sensitive attributes based on the relationship between non-sensitive attributes and sensitive attributes.The minimum set of attributes provides an effective method for selecting the candidate set of the following two privacy protection algorithms.(2)For the small size of non-sensitive attribute set,a differential privacy protection method based on non-sensitive attribute random response is proposed.The algorithm first calculatesthe non-sensitive attribute candidate set A by NS-SC algorithm,and then randomly responds non-sensitive attributes in candidate set A.(3)For the small size of sensitive attribute set,a differential privacy protection method based on sensitive attribute random response is proposed.The algorithm first calculates the non-sensitive attribute candidate set A by NS-SC algorithm,and then randomly responds the sensitive attribute,and finally the corresponding non-sensitive attribute is selected in A by the associated query.The result is the disturbed non-sensitive attribute.(4)The NSRR-DP and SRR-DP algorithms are analyzed by experiments,and the correctness and effectiveness of the algorithm are verified on different datasets.