Towards Generating Policy-compliant Datasets

Abstract

The development of intelligent (e.g., AI-based) applications increasingly requires governance models and processes, as financial legal sanctions are more and more being associated with violation of policies. We propose an ontology representing the informed consent that was collected by an organization and argue how it can be used to assess a dataset prior its use in any type of data processing activities. We demonstrate the utility of our ontology using a particular scenario, where datasets are generated “just in time” for a particular purpose such as sending newsletters. This scenario shows how data processing activities can be managed to in such a way as to support compliance verification. This paper furthermore compares the contributions to related work and positions it into prior work concerned with the broader problem of prescribing and analyzing compliance.