Research On Privacy Protection In Governmental Data Publishing And Sharing

With the arrival of big data era,government pays more and more attention to mining the value of government data by means of technology.At the same time,the demand of public for government data publishing and sharing is increasing.However,there are different degrees of privacy leakage risk in the government's data publishing and sharing.Therefore,how to publish and share government data under the premise of protecting citizens' privacy have become a challenging issue for improving government governance capabilities and service levels.For the requirements of privacy protection in government data publishing and sharing,this paper mainly completes the following research work:(1)This paper firstly analyzes the risk of citizen privacy in government data publishing and sharing.Then based on the government cloud + fog computing model,a privacy protection data publishing and sharing framework has been constructed.And the role and function of the cloud layers,fog layers,and user layers are define.Finally,the paper analyzes the advantage of cloud-fog computing model for information resource management,and designs the privacy protection government data publishing process based on fog layer and the government department data sharing process based on fog layer.(2)In order to solve the problem of privacy information leakage when attackers attacking the published data by using the greatest background knowledge.This paper propose a MaxDiff histogram data publishing algorithm,which is based on differential privacy(DP-based MDHP).Laplace noises are added to the original dataset by differential privacy method,,which can prevent citizens' privacy even if attackers get a strong background knowledge.According to the maximum frequency difference,the adjacent data bins are grouped,then the differential privacy histogram with minimum average error can be constructed.Through the application example,the application method of the algorithm is described,which shows that the published privacy protection histogram can meet the citizen's need for understanding the government statistical indicator data.Further,the DP-based MDHP algorithm and the LP algorithm were compared using the real dataset of the Filipino family income and expenditure provided by the Kaggle platform.Through the two indicators of absolute error and Kullback–Leibler divergence,the advantages and usability of the proposed algorithm are verified.(3)From the point of view of privacy protection requirements for data sharing among government departments,according to the characteristics of government data types and complex attributes,this paper proposes an improved K-anonymity data sharing method based on KMedoid clustering(KMedoid-based KADS).First,in order to reduce the sharing range of non-demand attribute data,the original data table is pre-processed and divided into multiple data tables based on attribute correlation.then,the similarity clustering algorithm based on KMedoid clustering is used to deal with the partition.The resulting shows that the data table initially meets the anonymity requirement.Finally,according to the sharing requirements of the data resource request department,an anonymity cluster sharing algorithm is used to provide an anonymous data table for the demand data.Compared with the Incognito algorithm,it is proved that the KMedoid-based KADS algorithm can effectively reduce information loss and improve the usability of shared data.