Privacy Preserving Data Mining Research In Outsourced Data Service Mode

With the development of cloud computing,cloud service providers provide increasingly powerful and complete data storage and analysis services.The data storage and analysis outsourcing model has been accepted by users.The cloud service provider's data outsourcing services include data storage services and data analysis and mining services.The outsourcing service model of data as a service has become more and more mature,but at the same time there is a huge security risk in the data outsourcing service outsourcing model.How to ensure the security of user data in the data analysis process under the outsourced data service model is one of the issues that need to be solved urgently.Most of the existing research is based on a trusted third-party assumption that the cloud service provider is trustworthy,but in reality the cloud service provider is not trustworthy.Concerns that privacy cannot be guaranteed after the user outsources personal data analysis to the cloud service provider hinder the development of the data-as-a-service model.The untrustworthiness of the cloud service provider is mainly caused by two factors:1.The existing privacy protection technology of the cloud service provider cannot meet the user privacy protection requirement in the data analysis process.For example,Facebook's user data breach during the data analysis that broke out in 2018.2.The database administrator has full data access rights.The security assumptions in this paper are untrustworthy for cloud service providers and based on this research.Privacy-protected data mining mainly includes:1)Privacy-protected cluster mining For existing privacy protection clustering mining algorithms can not balance performance and security issues,a cluster-based privacy protection algorithm is proposed.Good balance between the security and mining performance of the original data.The experiment confirmed the validity of the method through real data sets and simulated data sets.2)Privacy-protected association rules mining for new security threats in the data-as-a-service model under cloud computing environments,dedicated to solving multi-source data global association rule mining problems,based on FP-tree for multivariate data fusion,while using differential privacy for each The data source FP-tree protects privacy and completely solves the problem of untrustworthy cloud service providers under the data-as-a-service model.3)Classified mining of privacy protection.This topic addresses the new security issues in the cloud computing environment.Under the assumption that the cloud service provider is not trustworthy,data protection for privacy protection is conducted,taking into account both performance and accuracy.The innovations of this topic mainly include:1)Existing privacy protection data mining research is mostly directed at distributed deployment applications.This paper addresses the new challenges posed by untrusted cloud service providers in the cloud computing environment.This paper addresses the issue of privacy protection security mining.2)The existing methods generally adopt the encryption method.In view of the performance problems caused by the inherent computational overhead of the encryption method,this paper proposes a corresponding solution with small computational overhead.3)Due to the lack of precision in the existing algorithms due to data generalization and other operations,this paper proposes a solution that takes into account both performance and accuracy,on the premise of ensuring security.4)The existing method uses local data for mining,and there is a problem of unilateral mining results.This paper proposes a global multi-source data integration trusted mining method.In summary,there is a good research value in the academic field and a wide application prospect in the industrial field.The solution is innovative.