Privacy-Preserving Data Publishing in Linked Data Mashup Architectures

The mashup of microdata sources to form a data hub must fulfill a set of privacy preservation anonymity requirements that hinder data analysts to figure out sensitive information of the source datasets. This is relevant in a number of fields that include smart cities, electronic healthcare records and others. Linked data publishing architectures are not designed to adapt well to the requirements of existing approaches to sanitize the linked datasets, which do not always exploit the potential of semantics. Besides, the sanitizing protocols are not always controlled by a central coordinator. We propose a classification framework to decide on the distribution of control and partitioning of the dataset information models. Based on the framework, we define an approach to engineer privacy-preserving linked data mashups that defines the essential functionalities of privacy-preserving linked data publishing architectures. The classification framework and engineering method for data privacy preservation can have an implication for big data systems and emergent blockchain-based distributed ledgers.