Reseach And Implementation Of PINQ-based Differential Privacy Data Publishing System

With more and more products and services built around user data,the age of big data has brought personalized services and intelligent lifestyles.However,the privacy of users will inevitably be leaked in the process of data collecting,analyzing and publishing.As a new type of privacy protection method,differential privacy can not only resist arbitrary background knowledge attacks,but also prove its level of privacy protection with rigorous and efficient mathematical methods.It is currently a research hotspot in the field of privacy protection.As the first data analysis platform combined with differential privacy,PINQ platform can provide a strong security guarantee for the underlying data set.Therefore,this paper selects PINQ platform for research,designs and implements a PINQ-based differential privacy data publishing system,aiming at solving the balance between data publishing and privacy protection.First,this paper provides an overview of the background of differential privacy,research status,and related methods in the field of privacy protection.Then,it elaborates the related concepts and properties of differential privacy,and analyzes the main mechanisms of differential privacy in detail,such as Laplace Mechanism and Exponential Mechanism,after which the platforms and frameworks applying differential privacy are summarized.Second,this paper conducts an analysis of the source code of PINQ platform,gives a comprehensive introduction of the implementation of each aggregation method,and details its main function modules and calling relationship.Next,it summarizes the defects of PINQ platform and proposes the corresponding solutions.Since the original method,NoisyAverage,provided in the platform cannot guarantee complete differential privacy,this paper improves its source code to make up for this defect.The Exponential Mechanism is further applied in this paper to design a mode method,NoisyMode,that satisfies differential privacy,to improve aggregation methods of the platform.Finally,this paper designs and implements the PINQ-based differential privacy data publishing system.Through demand analysis and outline design,the system is divided into four modules:user management,data set management,data processing,and data publishing.After detailed design of each module,the system is completed by using SpringBoot+Vue.js technology stack.As such,this paper aims to break the barriers between data sharing and analyzing through the system,and consequently provides a secure data sharing and analyzing platform.