AitF: Provenance with Privacy and Reliability in Federated Distributed Systems

University Of Pennsylvania, Philadelphia PA

Investigators

Abstract

The goal of this project is to investigate the question of provenance in federated distributed systems, such as networks and scientific workflows, where several self-interested entities generate and share data, and compute and make decisions on the shared data. Computations and decisions give rise themselves to new shared data. Provenance analysis allows us to understand the contributions of the various entities to the data. This project seeks to put provenance on a sound mathematical foundation by unifying the theoretical notion of semiring provenance with practical approaches to network provenance. The PIs would also like to design new algorithms and compression techniques for collecting and managing provenance data. A second thrust of the project is to compute and associate reliability scores to each data item, in conjunction with their provenance. In order to do this, the PIs will design techniques for assigning reliability scores to primitive data items, as well as a calculus based on sound axiomatic principles for assigning reliability scores to derived data items. This project will achieve broad impact by allowing for networks to operate more reliably and by enhancing reproducibility in scientific workflows.