Research On Privacy-preserving Data Lineage Publishing

As data lineage becomes more widely used,the need to share data lineage is growing.The data lineage contains the source data and the evolution process information from source data to target data,sharing and publishing it inevitably brings the data lineage privacy protection problem.The lineage workflow is the main manifestation of the data lineage.It is one of the research hotspots to protect private information of lineage workflow while publishing the lineage workflow.In view of the shortcomings of the module privacy and structural privacy protection methods existed for lineage workflow,this thesis proposes methods to protect the module privacy and structural privacy of the lineage workflow.The thesis work is as follows:(1)There is a lack of attention to the privacy security impact of interaction between the module input and output attribute set when setting the hidden attribute set and the poor operability for the ?-privacy model.To solve problems above,the definition of the local mapping relationship and the local mapping set are introduced,and N?-privacy model is proposed from the perspective of destroying the local mapping relationship of modules.The heuristic method is designed to calculate the hidden attribute set,the method takes into account the impact of module sensitive attribute set and local mapping set on the module privacy protection.Besides,a ? setting strategy based on hidden attribute set is proposed.(2)There are some problems of weak theoretical foundation,unable to measure privacy protection effect quantitatively and weak maintenance of key paths of lineage workflow in the existing structural privacy protection method based on restricted publication of lineage workflow.This thesis proposes a privacy-preserving workflow publishing method based on differential privacy.The definition of key path and key path priority are introduced,on basis of this,the ?-Project projection algorithm is proposed to reduce the degree of the lineage workflow,and at the same time,the high priority key path reaching is maintained according to the preference of users for the key path.The concept of oi-sequence is introduced to extract the structure characteristics of the lineage workflow,and Laplace noise is added to the oi-sequence to satisfy the differential privacy model constraint.The ?-Project algorithm is used to adjust the global sensitivity of the oi-sequence after noise addition to reduce the Laplacian noise scale.Finally,the perturbed oi-sequence is used to reconstruct the lineage workflow for publishing,which realizes the workflow privacy security and the maintenance of key paths accessibility.Theoretical analysis and experimental results verify the effectiveness of proposed methods.