Research And Implement Of Privacy-preserving Scheme Based On Data Mining

With the rapid development of computer running speed,information technology and Internet applications,more and more information is being stored in digital form uniformly.The digital data can be effectively analyzed and processed by data mining technology,and their values also can be found out.The results of these data are able to help users make decisions and understand information easily.However,in many professional fields,such as finance,telecommunications,medical and other industries,the data are generated by data mining techniques may leak personal or business confidential information.How to prevent the loss of privacy and carry out data mining work with high accuracy became a issue to be solved.Therefore,the research of privacy-preserving based on data mining program has become one of the common research priorities in both data mining and information security field.Since access to the data will also obtain the user's personal information,which pertains to the user's personal privacy,protecting the user's information security and privacy of data without affecting the data mining results have great practical significance.This paper proposed a privacy-preserving scheme which is established on the basis of study of the theory of privacy,summarized the results of the research of existing data mining for privacy algorithm.To solve the problems that the existing k-anonymity algorithm cannot resist the homogeneous attack and high information loss and other issues,the optimization algorithm is utilized.Based on this algorithm and detailed system design and implementation,the paper evaluated the privacy effects and performance of the system.The test results showed that the system can efficiently protect the privacy of user under the condition of data publishing functions and small computational cost.The main contents of this paper are as follows:1.Firstly,the concepts of privacy,protection of privacy and data mining are introduced.Secondly,three main types of data mining algorithms were introduced.Thirdly,four existing privacy-preserving based on data mining methods are described,the advantages and disadvantages of each method are analyzed in detail.2.Combine with the characteristics of privacy issues,this paper made a deep research and analyzed for k-anonymity,proposed a improved scheme.Firstly,for reduce the loss of data,this paper clustered algorithm,and used k-anonymity algorithm for k operation.Secondly,to solve resist against attacks homogeneous that the original algorithm cannot addressed,the paper constrained the selected single sensitive valued attribute.Thirdly,it also improved algorithm based on a detailed scheme for the protection of the privacy of data mining,and performed detailed data distribution system of the overall design.3.The detail implementation of the data distribution system is shown in this paper.Firstly,implementation notes and pseudo-code of the user interface module,data import module,data preprocessing module,data mining module,a privacy protection module,the log management module and data distribution visualization module were presented.Secondly,the relationships between the various modules of the system were described.Thirdly,the improved algorithm performance and the effectiveness of privacy-preserving of the system had been tested.Finally,the paper summarizes the prospects and future work.