A Comprehensive Sanitization Approach for Workflow Provenance Graphs

As the number of provenance aware organizations increases, particularly in workflow scientific domains, sharing provenance data becomes a necessity. Meanwhile, scientists wish to share their scientific results without sacrificing privacy, neither directly through illegal authorizations nor indirectly through illegal inferences. Nevertheless, current work in workflow provenance sanitizing approaches do not address the disclosure problem of sensitive information through inferences. In this paper, we propose a comprehensive workflow provenance sanitization approach called ProvS that maximize both graph utility and privacy with respect to the influence of various workflow constraints. Experimental results show the effectiveness of ProvS through testing it on a graph-based system implementation.