Research And Implementation Of K-Anonymous Technology Based On Sensitivity Grading

In recent years,with the rapid development of information network technology,a huge amount of data has been produced in people's production and life.In order to meet the needs of data sharing,decision support,scientific research,etc.,data needs to be released.However,the data may contain individual private information,which may result in a privacy breach if it is directly posted or posted improperly.Therefore,how to avoid the privacy leakage caused by data release is a hot research topic in academia and industry.Anonymization technology is one of the research focuses.It can balance the availability and security of data,so it has received extensive attention and research in the academic community.The k-anonymity model is also the focus of this paper.Aiming at the problem of skewness attack and sensitivityattack in existing clustering-based(k,l)-anonymity algorithms,this paper combines them with(l,c)-diversity model based on sensitivity classification,and proposes a(k,li,c)-anonymous algorithm based on sensitive value constraints,which can resist multiple attacks and enhance the protection of highly sensitive attributes.However,the(k,li,c)algorithm uses uniform sensitive attribute frequency constraint parameters,which fails to make full use of the sensitivity grading.For this problem,this paper associates the sensitive attribute frequency constraint parameters with the sensitivity level,and proposes a(k,li,ci)-anonymous algorithm based on sensitive value constraints,which can effectively reduces the information loss during the anonymization process.This paper uses the Adult data set in the UCI database for simulation experiments.Experimental results show that the improved algorithm can effectively reduce the information loss in the process of data anonymization and improve the quality of anonymously published data.Finally,using the proposed anonymous algorithm,a privacy data publishing system based on SSH framework is designed and implemented.