Towards Fine-Grained Data Access Control Through Active Peer Probing.

Data is routinely being shared online by peers, for instance in business transactions, social activities and others. This data, in turn, is often transferred, processed and combined through complex querying and analytics. This raises questions such as the following: who owns the derived data? With whom and for what purpose may it be published? If consent is required for its dissemination, whose consent should be obtained? The related topics of data sharing, privacy and access control have been extensively studied, but uniquely our focus here is not on data management with known policies but rather on the active probing of peers to ask for their consent. Active probing has the potential to allow finer-grained access control, where it is unreasonable to expect data owners to publish their full policies, defined for all possible sharing scenarios. They may not even have a clear view of their own policies, before asked whether they are willing to share data with a specific third party. This short paper informally introduces and motivates this new problem. It further identifies interesting connections to two distinct areas: data provenance, which captures the way output data are derived from inputs, and Boolean evaluation, which focuses on effective strategies to probe hidden Boolean values for evaluating a formula. As we shall demonstrate, the composition of these two areas in the context of this problem yields intriguing avenues for further research.