Research On Low Concealment Rate Release Technology Of Multi-sensitive Attribute Data Based On Decomposition Method

With the advent of the era of big data,the issue of personal privacy data leakage has become increasingly serious.In order to ensure the availability of published data and protect sensitive attribute data from disclosure,research methods using multi-sensitive attribute data protection have become a hotspot in the field of data protection.The work done in this paper is to study the privacy data protection of multi-sensitive attributes.The MSB of multi-dimensional bucket grouping technology often produces high concealment rate and large information loss,analyzes the causes of defects,and designs improved algorithms.A multi-sensitive Attribute Data Maximum Selection Priority(MSADMSP)based on decomposition method is proposed.The algorithm first analyzes the association relationship between multi-sensitive attributes,divides the sensitive attributes into m sensitive attribute groups,reduces the sensitive attribute dimensions,and then adopts the selection strategy in the maximum multi-dimensional capacityfirst algorithm MMDCF according to the selected diversity parameter L.Selecting the appropriate grouping of the data sub-tables formed by the m groups to form an equivalence class group satisfying the diversity parameter l,and generalizing the content of all QI attributes corresponding to the group records capable of publishing the equivalence class,Then,for the published SA-sensitive attribute table,each SA sensitive attribute table randomly exchanges the sensitive attribute values contained in each equivalence class in units of equivalence classes,further disrupting the association relationship between the published data.Finally,the raw data is decomposed into a QI table and m SA tables and published using lossy link technology.This method effectively reduces the concealment rate of published data,weakens the relationship between sensitive attribute data,and achieves a dynamic balance of data availability and security.