Computing Science: Preserving Privacy in Shared Provenance Data.

摘要

Provenance management still lacks robust models for sharing provenance data between multiple parties while keeping parts of it private to the owner. This limits the potential for provenance dissemination, which is a critical step in enabling data sharing amongst partners with limited a priori mutual trust. In turn, this has a negative impact on data-intensive science and its associated research publication repositories, on audit tasks, as well as on increasingly common collaborative dynamic coalitions scenarios. We propose a method for preserving privacy by creating abstractions over provenance graphs, we apply it to provenance sharing, and illustrate it on a health care case study.