Research On Privacy-preserving Provenance Workflow Publishing

As data sharing continues to deepen,the need to share and publish Provenance Workflow,which describes the principles of data generation and evolution,is increasingly pressing.Provenance workflow has important application value in tracking historical information,data recovery,data source citation,etc.Directly sharing provenance workflow has the risk of leaking workflow privacy.The issue of privacy protection in provenance workflow sharing has become a hot topic for researchers.In view of the shortcomings of the existing provenance workflow module privacy and structural privacy protection methods,the privacy-preserving provenance workflow publishing method that maintains the traceability query and the privacy-preserving provenance workflow publishing method that maintains the critical path available are proposed.The main work of the thesis is as follows:(1)Existing module privacy-preserving methods split the relationship between the module and the workflow structure,failing to account for the importance of the module in the data evolution process,resulting in poor availability of the traceability query of the published provenance workflow.Aiming at this issue,a method for privacy-preserving provenance workflow publishing is proposed,which maintains traceable query availability.By collecting a large number of module participation samples during the random execution of workflow,a Bayesian network model is constructed to measure the degree of dependence between related modules in the workflow,thereby determining the role of different privacy modules in workflow traceability query.A privacy protection method for personalized modules is proposed.Based on the constructed Bayesian network,the strong and weak correlation modules of the workflow are divided,different hiding strategies are designed.The hidden processing of the privacy module is maintained locally to reduce the modification of the workflow structure to maintain the availability of traceability queries.(2)Anonymity-based workflow path privacy protection methods do not pay attention to hiding the true length of the path,and when the number of paths between the target nodes is lower than the anonymous strength k,the weight perturbation method that maintains the graph structure unchanged cannot meet privacy protection claim.Aiming at this issue,a privacypreserving provenance workflow publishing method is proposed,which maintains the critical path availability.A workflow(k,?)-critical path anonymous privacy-preserving model is introduced to ensure that the critical path between target modules in the workflow meets kanonymity.An anonymous method for critical path of provenance workflow is proposed.By combining weight perturbation and module decomposition strategy,(k,?)-critical path anonymity can be achieved in different scenarios.While trying to avoid adding fake information to the structure of workflow graphs,it prevents privacy attacks based on critical paths.Theoretical analysis and experimental results show that the proposed method can effectively prevent the leakage of lineage workflow privacy while effectively maintaining the data availability of specific applications in the workflow.